【突發】馬斯克、黃仁勳同台宣佈：AI不停了，還要加速前進！美國必須贏！美國要用「超級智慧」徹底甩開中國！馬斯克當眾調侃：黃仁勳一個人就創造了5兆美元財富！【中英字幕】

0:00 If you advance capability, it somehow compromises safety. If you advance safety, somehow you fall behind. That that's just that, you know, those are false choices. >> I think essentially China is the biggest competitor and and China has um about three times the electricity production of the United States. AI is just an incredibly helpful, useful thing. Um, you know, I've heard many anecdotes of AI helping people uh with medical issues. Um, and in in many cases where the the doctor was wrong and the they they submitted their X-rays or MRI to to AI and and AI got it right and and but but for AI, they would have died. >> On behalf of all Americans, I want to thank both of you. I think combined you

0:46 have created more than5 trillion dollars of wealth for ordinary Americans in their 401ks, tens of thousands, maybe hundreds of thousands of jobs directly, many tens of billions of wages. So, thank you. Um, [applause] I think Jensen's done five trillion by himself actually. >> I'm sorry. >> I think Jensen's done five trillion by himself. [laughter] >> I think I think I think around three and a half or something. Yeah, I was trying I was trying to be conservative. I'm an ordinary American and uh I use I use um uh Clyde to uh help me do research and and um instead of giving me just answers and most of my research that maybe there's comparative

1:32 research that I'm doing and there's two or three different approaches to something and I wanted to answer a list of different questions and uh compare and contrast uh rates of rates of growth or rates of change or whatever it is and and so I describe literally, you know, every single question I have and nothing gives me more joy than and then and I ask it to create a nice report and and plot it out and sometimes it's a pie chart, sometimes it's a bar graph and I just hit enter and nothing gives me more joy than just watch it go and it comes back and it creates a beautiful, you know, beautiful report and and then instead of instead of researching all these different topics for four, six, eight hours, which back in the old days I would on a weekend, you know, learn something new, it goes off and creates me this really really beautiful final

2:20 report in about 3 minutes and you know it's all condensed and now you're getting smarter faster. >> Hey team. >> Oh gee. >> Are we late [laughter] guys? >> Good to be here. >> Go America. >> Go America. Team America >> go. >> Yes indeed. So America Wright brothers first man on the moon. the internet. >> The two of us, right, brothers. >> I like that indeed. We've we've essentially led every major technological revolution which has been to the enormous benefit of America, ordinary Americans. We're I guess we're now calling this super intelligence, which I think is actually good branding. And so, let's let's talk about how

3:05 America wins this and we make that winning good for ordinary Americans. Well, um, excuse me, Diet Coke. Sorry. [clears throat] Uh so first of all artificial super intelligence super intelligence is a a complete reinvention of the computing stack. Uh everything from the way the processors the chips are are designed and the way they operate to the way software is uh now trained and and developed to the way that the applications are able to perform what they're able to do and how they how they perform. every single layer of the computing stack has been reinvented. Um, this way of doing computing unlike previous ways of retrieving information,

3:51 this is a generative approach to to computing and therefore in order for us to succeed the first thing we need is energy because we need the energies to to run the compute. the compute run the algorithms, the algorithms um then feed the applications and so at the at the foundational layer is energy and uh from that energy we build the land power and shell and that host the infrastructure so on so forth and so the way we win um of course we are world class in chip design we're world class in algorithm development um and uh uh and so one of the things that we need to do is just be vigilant about enabling power delivery and power generation to happen. And President Trump uh has been has been visionary in being pro- energy. Without

4:37 the pro-eny vision of President Trump, it would have been impossible for us to build the data centers on top of it. Um now, of course, uh these data centers are coming online so quickly. Uh we're we're uh uh coming into a lot of neighborhoods and a lot of communities. We have to do a better job um in partnering with the communities and and to make and bring them along and make them part of the prosperity and the success of this industrial revolution. Um but you know that's how we win. You know we got to build the world's best chips. We got to build it. We have to have lots and lots of energy. Uh create incredible algorithms. And then lastly, make sure that America is enthusiastic about the technology so that we diffuse it into all of these various industries from manufacturing to healthcare, you know, space uh space exploration, so on so forth.

5:23 >> Awesome. Elon, how do we win? How do we keep winning? Well, [clears throat] we need um a lot of power, a lot of energy uh generation. So, uh scaling energy, I think, is is one of the biggest things. You've got to scale energy, you've got to scale chip production. Um and I I think we're we're winning on software on on anything intellectual or digital. Um but uh long in the long run we need to be able to generate enough power uh to compete with uh I think essentially China is the biggest competitor and and chi China has um about three times the electricity production of the United States. Um and

6:10 I think we'll figure out how to make shift. So, so the the the really the central challenge I think is long long term is going to be how do you generate enough power and how do you ensure that uh we have enough uh logic and memory fabrication capacity in the United States or in a safe region. >> Well, you are >> we need the right brothers on the job. >> Yes. Well, turns out the right brothers are on the job. Elon, if you bring on 10 gigawatts, which SpaceX has has discussed publicly, that'll be about 25% of all the power added in America. And I know you have literal rocket engineers working on bringing on the power. >> Yeah. >> Helping make the turbines. So, can you can you talk about that? >> Yeah, it's I mean, just to to frame

6:56 this, um the average power consumption in the United States is about 500 gaw. So every 5 GW of incremental steadyst state production is a 1% increase in in the power used. Um and and I think actually will materialize to be approximately 1% increase in GDP. Um because the the intelligence for what keeps increasing. So if you just you know plot on a curve um how is how is the useful intelligence for what increasing? um how is the both at the hardware level Jensen's uh GPUs keep getting better in terms of their the you the compute per watt um and the uh algorithms also get better in using and getting more intelligence per watt so um

7:42 so I that's a maybe an interesting note I I I would bet anyone that 1% increase in power usage corresponds to roughly 1% increase in GDP and and so 10 10 gawatt would be a 2% increase in in GDP um with the Reubens I think. Um so >> well let's do let's do the simple math. So each gigawatt um can probably generate about call it 4050 billion dollars of economic output each year. >> Uh yeah I mean 1% would be yeah I mean >> 1% of the United States >> would be >> 300 billion. >> Yeah. 300 billion and so 506 >> what 1% of the of the 1% increase so 5 gawatt would be yeah well 5 gaw would be

8:28 300 billion exactly 60 60 billion g that's right >> the math maths >> yeah that's what we're good at >> yes >> so so that's that's quite >> not good at too many other things but me >> um yeah I mean it's it's going to increase quite dramatically year-over-year. So, yeah, I mean, and and from a SpaceX standpoint, we're looking at long-term, how do we access a lot more energy than could be accessed on the ground? And and that's with space. So, uh orbital compute is going to be a very big deal. Um in fact, and Google agrees. They they just actually sent up uh or either about to or soon

9:15 soon about to send up some TPUs. I think there's that there was a um an H100 that was put into orbit like last year, I think. >> Mhm. >> Um so it it does work in in space and um and and SpaceX is aiming together with Tesla to do 200 gawatt of solar production per year. And if brought to space that you get the name plate or better on on the solar power whereas if it's on the ground you're going to get somewhere between a fifth and an eighth of the uh solar power generation and you need enormous batteries. So whereas it's uh essentially always sunny in space. So uh and right now is a great time to do it because the economic returns of these computing systems are so good. Yeah.

10:01 that the the launch cost although is still quite high um in terms of the overall economic benefit is quite quite affordable. >> Absolutely. >> This is the perfect time to do it. Putting Nvidia compute in space is the best way to get there. >> Yeah. Climbing the Carter scale with American power and American chips on American rockets. I love it. >> Yeah. So Jensen, you said something that I thought was important about how when we bring the energy to all these small towns, we're reindustrializing America. We're revitalizing small towns. We're modernizing the grid. These are things that essentially every American politician has wanted for the last 35 years. What would you say? What what what are some specifics you can give

10:47 about how data centers are good for ordinary Americans in the communities and how they're getting better for those communities? >> Yeah, we used to call them data centers. We call them data centers because that's where we hold data. But the things that we're building now, these are really SI factories, super intelligence factories. And the reason why it's a factory is because you're producing something of economic value. You're just not storing it. And so so uh these SI factories um are real factories. It's like a you know big giant power generator in real factories. It's got a lot of computing in there, networking in there, power generation, cooling systems. It's complicated things. And so we're building out, you know, just as the internet today allows you to find anything anywhere in the future with these AI, excuse me, SI infrastructure,

11:33 SI factories, you'll be able to know anything and do anything from anywhere. And so uh now we're building this infrastructure out and you know we call it uh across all of the United States uh somewhere between 10 uh 20 gigawatts a year. Let's just just estimate that creates just you know probably about a million jobs in in uh building out these data centers. Don't forget you've got to build the power generation plants, you got to build the data centers, you've got construction, you've got you know power cooling, you've got you know pipe fitters. And so this is the first time in probably as you mentioned 35 I was going to say 50 years first time in 50 years we are re-industrializing the United States creating a whole bunch of blueco collar jobs and um I and I think

12:19 we want we want an econ we want an economy that has a shape that includes white collar jobs and blue collar jobs not just white collar jobs and so so I think that this is our great chance of reshaping our economy and bringing that type builders and craft backed into the United States. >> Absolutely. And it I mean it feels like what uh what Colossus and Macro hard and macro harder have done have been amazing for the communities around them, the jobs they've created. How do you how do you think about that? >> Uh yeah, the I mean the economic impact is positive economic impact has been tremendous. Um I mean we we there's actually a job shortage in in the region and whereas there used to be uh quite a bit of unemployment and now it's uh

13:05 there there's a overmployment if you will there's there's more jobs than than there are people um and the the tax benefits uh for the community have been tremendous. I think the I think we might have doubled their their tax budget or something um quite quite tremendous. Then we we do like small things that are helpful like provide you know low price halfp price stallink uh access. Um we're building a water recycling plant that's a $250 million water recycling plant. Um I think you you do need to do things for the community. Um and if you do so they will reciprocate. >> Awesome. Now one of the the things that people people forget is that for the first time in in probably forever that we have the market forces to support the investment of sustainable energy. You

13:52 give me an example of a fusion fision you know small nuclear reactor large reactor almost every form of sustainable energy is being funded today. And so we have the opportunity to fund the future of energy without without the government subsidizing without right. I mean we this is you give me an example of a great energy battery storage idea. It's going to get funded. Awesome. By by the way on behalf of all Americans I want to thank both of you. I think combined you have created more than5 trillion dollars of wealth for ordinary Americans in their 401ks. Tens of thousands maybe hundreds of thousands of jobs directly. Many tens of billions of wages. So, thank you. Um, [applause]

14:39 >> I think Jensen has done five trillion by himself actually. >> I'm sorry. >> I think Jensen's done five trillion by himself. [laughter] >> I think I think I think around three and a half or something. >> Yeah. >> I was trying I was trying to be conservative. Yes. >> So, you guys both had a great Monday. I was watching the uh Starship launch. first orbital Starship launch while I was reading about the open agent safety framework from Nvidia. Uh which was an exciting morning. Uh so may maybe you could each talk about each of those because we need Starship to get those 200 gawatts of American solar with American chips, American super intelligence into orbit. >> I'll talk about his talks on mine.

15:24 >> And so no, I'm just kidding. So So um uh let's see. uh as you know uh we want to advance AI advance SI we I'm going to get good at this we're going to advance SI in a in a responsible uh and optimistic way and we we know that the future uh we strongly believe that the future that SI is going to enable is a future that we want to be part of and so we're optimistic about that and that's the reason why we're driving so hard on the other hand we have to make sure that we develop this technology safely and securely there's several different parts of it while you're training the model. Um, that's that's generally a fairly safe environment. But while you're evaluating the model, while you're testing the model, you have to make sure that the agent agentic system, the model itself is highly contained. That

16:11 containment system has to be really secure. And the second thing is that while you're while you're testing it, you want to monitor that that just because just even though even though you have it contained, you should never trust that it's contained. And so you want to monitor it in real time. And you want to ideally monitor it with another another chip so that it is outside, you know, if you will call out of band, outside the environment, outside the sandbox of the agent itself. And so you want to contain and you want to monitor. Um once you're done uh evaluating in this way, when you deploy, you want to use a very similar system. And this is no different than than um uh just about any digital any human employee that you hire. uh you give them a containment of rights and tools and access uh and then

16:57 you monitor on on a regular basis on a continuous basis. And so in the case of agents, we just have to make sure containment and monitoring has to be done in the most secure of ways. And we in introduced two technologies. One's called open shell. Uh it's a a containment system. Uh it's basically kind of like a browser for websites. It's now essentially a browser of agents. Um, and that browser keeps uh the agent isolated from the rest of the computer and the rights that you give it are only the rights it should have. And then the the the monitoring system uh is a a new chip called Bluefield and it it just keeps an eye on everything that the agent is doing at all times. And if it does anything that's you know out of bounds and against the policy that has been given uh then it you know escalates

17:45 uh as far as you need it to escalate. And I thought it was a great analogy u in in the technical blog where he said in the early days of internet everybody was worried about your credit card could get stolen. We made the internet safe and that's why we have Amazon, we have Netflix, we have Google, we have X, we have so many so many great things because we made the internet safe. Now we're do now now we're doing the same with AI >> and safety is not you know inconsistent with advancing the technology. Safety and advancing a technology is one and the same thing. is no different than than um uh when you have autopilot. Not only is it advanced technology, it's also safer technology. And so you want to you advance the technology in that way. Um the two are not in conflict. Um the idea that if you advance capability,

18:30 it somehow compromises safety. If you have man safety, somehow you fall behind. That that's just that, you know, those are false choices. >> Yeah. Safety lets us accelerate. But Elon, if the Starship's going to be a big part of those 200 gigawatts, what can you tell us about Monday's flight? What you learned? How you feeling about the program? >> Yeah, I think the Starship is going great. Uh we we uh we took Starship to orbit for the first time and we deployed uh Starlink B3 satellites which are enormous satellites. I mean these are satellites with the wingspan of 737. So u really really enormous uh in fact it's it's the largest payload delivered to orbit since Skylab. So uh now we expect Starship to uh ultimately next year I think probably

19:18 reach um a weekly or twice weekly cadence. >> We've got not Tom Brown here. I know him by his exhandle. Welcome to >> good to see you guys again. [laughter] Um it >> was a long lunch. >> Yes, it was. Uh, my my car took a wrong turn, so you guys got here faster than me. Impressive. [laughter] >> Glad you made it. So, but Elon, please please continue. We're talking about how Starship is going to let us get to 200 gigs of American solar with American chips, with American super intelligence. >> Yeah. So, I mean, Starship is going to do a lot of things. It's I mean it's intended uh to carry enough payload to the moon and Mars to build a self-growing city on the moon or Mars. So um I mean I think as a rough order of magnitude you need about a million tons

20:04 to deliver to the surface of the moon to uh create a self-growing city or similarly similar number for Mars. Um and and and then in addition it will launch um at least a few hundred gawatts a year of uh AI compute. um 200 gigawatts a year would be a 40% increase annually in the United States total energy uh consumption. So that's quite a quite a big number. Um and it might end up being a terowatt one day. Um so this is this is serious tonnage. >> Awesome. And and Tom uh Enthropic uh your your company you're a co-ound founder of Enthropic. you announced Opus 5.5 in the last several days and I thought maybe we'd go to the exponential that we are on and I think it's helpful to ground what can Opus 5.5 and maybe

20:52 it's for everybody what can Opus 5.5 do to make the lives of ordinary Americans better today and where are we going to be in a year? Yeah. So, um, Opus 55 is, uh, a model that I think just continues the normal progress that we've had of getting kind of more IQ points for each of these super intelligences, uh, with each model release. Um, and so the main things that people will see if they use it now is that it can do a task that takes a longer time than they'd be able to delegate to to a previous model. Um, and so I think the things that people should see is just like as a normal American, if you want a task, if you want to actually delegate a task to it, it will be able to do that better is one thing. And then two is a thing that I

21:38 think people notice less, which is just that there's going to be a huge amount of software being written by these models. And so all of the products that people use are going to be getting higher quality, cheaper over time. And it'll have this effect of just like all of the all of the tools that we use now were written by like people doing engineering work and now it's going to be written by both people and super intelligence is doing much much more engineering work than was possible before. >> Absolutely. But I think it's doing so much more than this. There's already multiple stories of AI. I'm sorry. Super intelligence. super intelligence finding a life-saving drug that can be used for a disease that it wasn't understood that

22:26 it could safely treat. So, it's saving lives. I think it's having profound impacts on education, but may maybe each of you could talk for ordinary Americans. What's super intelligence doing today and what's it going to be able to do in a year? I'm an ordinary American and uh I use I use um uh Clyde to uh help me do research and and um instead of giving me just answers and most of my research that maybe there's comparative research that I'm doing and there's two or three different approaches to something and I wanted to answer a list of different questions and uh compare and contrast uh rates of rates of growth or rates of change or whatever it is and and so I describe I literally, you know, every single question I have and nothing

23:11 gives me more joy than and then and I ask it to create a nice report and and plot it out and sometimes it's a pie chart, sometimes it's a bar graph and I just hit enter and nothing gives me more joy than just watch it go and it comes back and it creates a beautiful, you know, beautiful report and and then instead of instead of researching all these different topics for four, six, eight hours, which back in the old days I would on a weekend, you know, learn something new, it goes off and creates me this really really beautiful final report in about 3 minutes and you know it's all condensed and now you're getting smarter faster. >> Love it. What about for you? How is how do you how are you seeing super intelligence helping ordinary Americans and what do

23:57 you think it'll do in a year or two? >> Maybe a year or two. Take a year or two from now. >> Yeah. Well, I I think it's it's helping people organize their lives and uh do better at their jobs. So, I think it really will become um essential if it isn't already for people to use AI uh for in their professional lives certainly and and in their private lives. And so, AI is just an incredibly helpful, useful thing. Um, you know, I've heard many anecdotes of AI helping people uh with medical issues. Um, and in in many cases where the the doctor was wrong and the they they submitted their X-rays or MRI to to AI and and AI got it right and and

24:44 but but for AI, they would have died. And I've heard many such cases. >> Awesome. And hopefully Opus 56 tells Tom, "Don't take that left, take the right." [laughter] >> Hopefully, [clears throat] >> I think it's important these these super intelligences, they are imbued with the values of their creators and we do have there there are American values that we all share. The first amendment, free speech, commitment to truth. Can you can you talk about what each of you are doing to imbue help make sure that these these super intelligences are imbued with with the right values? >> Are we allowed to say anything controversial? >> Say something controversial.

25:31 >> Okay. So, so um I the the SI are are tools and and the challenge with an SI with values uh meaning it has judgment. um depending on how you use that SI, it may decide uh um how it deci it may judge you and it's effectiveness would be affected by that judgment. And so it's a little bit like a steak knife where if you if you uh were served medium well, it cuts perfectly, but if you were served rare steak and it looks awfully like live meat, somehow it doesn't cut. And so I I think there's a there's a you know there's a there's a question in my mind

26:18 yet about how much judgment you really want these tools to have. And and so that that the intentions of the user um you know obviously I use I use the knife as an example but you can think of a whole bunch of other examples. Uh the intentions are a little bit different. you know whether you're a defender or an attacker for example in cyber security the defender's actions could resemble an attacker and you want the defender you want this SI to um uh activate on behalf of the defender even though it appears to be an you know an aggressive action. Great. I think we're running up on time but maybe if you guys could share you spent a lot of time at the White House. Is there

27:04 anything that that you can share share with the world about what came out of what was evidently a a uh a long and productive lunch? >> We did good. >> Yeah. >> We we signed a joint declaration uh regarding SI safety. Um and uh I think I think that'll be published by the White House soon if it hasn't been already. Uh but uh we we agreed to to a number of things that uh include joint monitoring um board special committees and uh just generally uh grading each other's homework which is a lot better than if people just grade their own homework uh for obvious reasons. Yeah. And and includes every it includes everything from um rigorous internal controls um wellestablished methods,

27:50 well established processes, um very clearly documented intentions of the products that you're building, internal controls to make sure that you're delivering on those intentions. Um internal audit um that that escalates uh external audits um from from a third party and uh um and then industries sharing best practices. I I thought that the the the declaration has has has a lot of teeth. Um I thought it was really well written and um I think this is a a practice that that quite frankly as a standards for our industry going forward um every every single company out of out of embrace. And so pus pus pulled it all together. Uh he he um watched over the meeting and uh uh was able to bring

28:35 everybody out and um you know I thought it was it was really wellrun. >> Awesome. Well, on behalf of all Americans, thank thank all of you for doing everything you can to make the future as awesome and optimistic for American Americans as you can and you are. So, thank you. [applause]