1 00:00:00,160 --> 00:00:04,477 Please, please have a seat. Now, I'm sure we can do a bigger round of 2 00:00:04,560 --> 00:00:07,037 applause for one of the greatest two 3 00:00:07,120 --> 00:00:11,277 leaders of our history. Let's go ahead. 4 00:00:11,360 --> 00:00:15,117 So, we're talking about I lost count, 5 00:00:15,200 --> 00:00:17,357 you know, 7 to8 trillion dollars worth 6 00:00:17,440 --> 00:00:21,597 of market cap comp. I lost count. But 7 00:00:21,680 --> 00:00:24,637 right now, we're here to celebrate a 8 00:00:24,720 --> 00:00:28,717 historic moment. A moment that yesterday 9 00:00:28,800 --> 00:00:30,957 during the dinner and thank you for for 10 00:00:31,040 --> 00:00:34,637 joining us under the patronage of the 11 00:00:34,720 --> 00:00:38,637 honorable president and his royal highness the crown prince Musedi where 12 00:00:38,720 --> 00:00:40,397 we had the pleasure to hear it 13 00:00:40,480 --> 00:00:44,317 firsthand. This is the greatest alliance 14 00:00:44,400 --> 00:00:47,357 between the Kingdom of Saudi Arabia and 15 00:00:47,440 --> 00:00:50,077 the United States where we have joined 16 00:00:50,160 --> 00:00:52,637 hands and you have helped us build our 17 00:00:52,720 --> 00:00:55,437 energy based economy fueling and 18 00:00:55,520 --> 00:00:58,237 energizing the industrial age and now 19 00:00:58,320 --> 00:01:00,397 fast forward going to the intelligence 20 00:01:00,480 --> 00:01:04,557 age where we can fuel AI factories 21 00:01:04,640 --> 00:01:08,557 robotics EVs and all of the rest. 22 00:01:08,640 --> 00:01:12,237 Speaking of that, let's start with you Elon if you don't mind Jensen and feel 23 00:01:12,320 --> 00:01:14,637 free to chive in you have a big 24 00:01:14,720 --> 00:01:17,597 fascination of something all of us have 25 00:01:17,680 --> 00:01:20,157 admired first order thinking which 26 00:01:20,240 --> 00:01:22,477 Jensen sometimes calls first order 27 00:01:22,560 --> 00:01:25,517 scaling which is an opportunity for how 28 00:01:25,600 --> 00:01:28,557 you have dropped the cost of batteries 29 00:01:28,640 --> 00:01:31,757 from a thousand for kilowatt hour to 30 00:01:31,840 --> 00:01:34,637 subund bucks and right now you're doing 31 00:01:34,720 --> 00:01:36,877 the same thing with robotics for 32 00:01:36,960 --> 00:01:40,877 actuators with servo rotors and motors. 33 00:01:40,960 --> 00:01:43,837 So I want to hear from you. How do you 34 00:01:43,920 --> 00:01:46,957 manage to always disrupt every single 35 00:01:47,040 --> 00:01:49,677 industry with that thinking? 36 00:01:49,760 --> 00:01:52,557 Well, it's mostly not disruption, it's 37 00:01:52,640 --> 00:01:58,317 uh creation. So with uh say SpaceX with 38 00:01:58,400 --> 00:02:00,397 reusable rockets, uh there really 39 00:02:00,480 --> 00:02:04,077 weren't any reusable rockets. Um, 40 00:02:04,160 --> 00:02:08,957 but the essence of getting of revolutionizing space travel is 41 00:02:09,040 --> 00:02:10,877 reusability. If you throw the rocket 42 00:02:10,960 --> 00:02:13,277 away every time, the cost of access to 43 00:02:13,360 --> 00:02:16,637 space is extremely high. Um, with 44 00:02:16,720 --> 00:02:21,197 respect to electric cars, there there weren't any electric cars when we 45 00:02:21,280 --> 00:02:25,997 started making them really. You couldn't buy any to the best of my knowledge. Um, 46 00:02:26,080 --> 00:02:27,917 so with Tesla, we wanted to make 47 00:02:28,000 --> 00:02:30,077 electric cars compelling um, and 48 00:02:30,160 --> 00:02:33,917 affordable. That was the goal. Um the uh 49 00:02:34,000 --> 00:02:38,157 you know with respect to humanoid robotics, there are no useful humanoid 50 00:02:38,240 --> 00:02:42,157 robotic robots at this point. Um there are sort of gimmicks, but there are 51 00:02:42,240 --> 00:02:47,037 there are no actually useful humanoid robots. Um and I think Tesla's going to 52 00:02:47,120 --> 00:02:51,037 make the first actually useful humanoid robots. Um and this will be quite a 53 00:02:51,120 --> 00:02:56,077 revolution and I think something that will that everyone will want. Uh because 54 00:02:56,160 --> 00:03:01,357 I always think of like who who wouldn't want their own personal C3PO R2-D2. 55 00:03:01,440 --> 00:03:05,037 >> Oh yeah, >> of course everyone would want one, 56 00:03:05,120 --> 00:03:09,917 right? And and then there would be many in industry uh providing products and 57 00:03:10,000 --> 00:03:13,357 services. This is why I say that humanoid robots will be the biggest 58 00:03:13,440 --> 00:03:15,917 industry or the biggest product ever. 59 00:03:16,000 --> 00:03:18,477 Um, bigger than cell phones or anything 60 00:03:18,560 --> 00:03:20,557 else because everyone's going to want 61 00:03:20,640 --> 00:03:23,437 one and uh or maybe more than one and 62 00:03:23,520 --> 00:03:26,877 there'll be many in industry. Um, 63 00:03:26,960 --> 00:03:30,557 >> I just want R2-D2 in C3PO's body. 64 00:03:30,640 --> 00:03:34,477 >> Yeah, there you go. Um, [laughter] >> well, I mean, a humanoid robot would be 65 00:03:34,560 --> 00:03:36,717 better than R2-D2 and C3 VO combined. 66 00:03:36,800 --> 00:03:37,917 Yeah. Times 10. 67 00:03:38,000 --> 00:03:41,917 >> Yeah. So the it and and you know people 68 00:03:42,000 --> 00:03:45,037 often talk about uh sort of eliminating 69 00:03:45,120 --> 00:03:49,037 poverty and that kind of thing but really the there's how long have they 70 00:03:49,120 --> 00:03:53,357 been talking about that? Um there's lots of talk uh you know there's lots of 71 00:03:53,440 --> 00:03:55,677 NOS's sort of trying to do these things 72 00:03:55,760 --> 00:03:58,717 but but really not succeeding. Um and 73 00:03:58,800 --> 00:04:00,877 and you know the evidence speaks for 74 00:04:00,960 --> 00:04:04,717 itself. Uh but but but AI and humanoid 75 00:04:04,800 --> 00:04:07,997 robots will actually eliminate poverty. 76 00:04:08,080 --> 00:04:10,717 And Tesla won't be the only one that makes them. I think Tesla will pioneer 77 00:04:10,800 --> 00:04:14,317 this. But there will be many other companies that make humanoid robots. But 78 00:04:14,400 --> 00:04:19,197 there there is only basically one way to uh make everyone wealthy and that is AI 79 00:04:19,280 --> 00:04:21,037 and robotics. 80 00:04:21,120 --> 00:04:23,277 And we can't talk about robotics without 81 00:04:23,360 --> 00:04:26,317 AI factories. And yesterday was such a 82 00:04:26,400 --> 00:04:29,197 historic day for the two nations but 83 00:04:29,280 --> 00:04:32,557 also for all of us where we celebrate 84 00:04:32,640 --> 00:04:35,677 the AI strategic partnership with the US 85 00:04:35,760 --> 00:04:38,317 s signed witnessed by the honorable 86 00:04:38,400 --> 00:04:40,797 president and his royal highness about 87 00:04:40,880 --> 00:04:44,797 how we are committing our capital energy 88 00:04:44,880 --> 00:04:49,677 land to energize the AI US ecosystem to 89 00:04:49,760 --> 00:04:54,557 be able to build inference node training nodes and to be the most AI enabled 90 00:04:54,640 --> 00:04:57,357 nation. With that announcement, tell me what's 91 00:04:57,440 --> 00:04:59,677 what's next in AI factories. Jensen, 92 00:04:59,760 --> 00:05:01,837 >> there there's a there's a beautiful 93 00:05:01,920 --> 00:05:04,397 story about how Saudi Arabia's building 94 00:05:04,480 --> 00:05:08,877 AI refineries now building AI fac or oil refineries to AI factories. 95 00:05:08,960 --> 00:05:12,717 >> I love that. >> I you know I've said that that AI is an 96 00:05:12,800 --> 00:05:15,997 infrastructure and the reason for that of course we understand AI from the 97 00:05:16,080 --> 00:05:20,557 perspective of the technology and how it's revolutionizing every industry. 98 00:05:20,640 --> 00:05:22,477 Digital intelligence of course has 99 00:05:22,560 --> 00:05:26,077 applications into every every field and 100 00:05:26,160 --> 00:05:27,677 so it's going to be used by every 101 00:05:27,760 --> 00:05:30,237 company, every industry, every country. 102 00:05:30,320 --> 00:05:33,597 In that way, it's foundational and 103 00:05:33,680 --> 00:05:35,357 therefore it's part of infrastructure. 104 00:05:35,440 --> 00:05:39,357 What is new about AI from a computer 105 00:05:39,440 --> 00:05:42,637 science perspective is that the way 106 00:05:42,720 --> 00:05:47,037 computing was done in the past was largely retrievalbased computing. 107 00:05:47,120 --> 00:05:49,597 Somebody typed in a story or somebody 108 00:05:49,680 --> 00:05:52,637 created a a piece of art or came came up 109 00:05:52,720 --> 00:05:55,517 with four versions of a digital ad or 110 00:05:55,600 --> 00:05:59,357 it's all pre-built by somebody which is 111 00:05:59,440 --> 00:06:02,157 then using a system to retrieve the 112 00:06:02,240 --> 00:06:03,677 appropriate version for you. It's a 113 00:06:03,760 --> 00:06:07,277 retrievalbased computing model. Hadoop 114 00:06:07,360 --> 00:06:12,157 and many of the the the frameworks and operating systems of the past all 115 00:06:12,240 --> 00:06:16,877 designed to retrieve the appropriate information for you. But today software 116 00:06:16,960 --> 00:06:19,117 is going to be generated in real time. 117 00:06:19,200 --> 00:06:23,277 It's generative based on the context, based on the circumstance, based on who 118 00:06:23,360 --> 00:06:27,117 you are, based on the problem you ask that based on your prompt. It will 119 00:06:27,200 --> 00:06:32,077 generate unique content for you every single time for everybody. It's unique. 120 00:06:32,160 --> 00:06:36,477 When you use Grock, every time you use it is different based on right based on 121 00:06:36,560 --> 00:06:40,397 based on the based on the prompt that you give it and based on the circumstance and 122 00:06:40,480 --> 00:06:43,037 >> and so therefore it used to be 123 00:06:43,120 --> 00:06:44,957 retrievalbased 124 00:06:45,040 --> 00:06:47,437 today it's generative and if it's 125 00:06:47,520 --> 00:06:52,077 generative then and every time is different then you need AI factories all 126 00:06:52,160 --> 00:06:55,757 over the world to generate the content in real time which is the reason why you 127 00:06:55,840 --> 00:06:59,757 need AI factories and and this is a unique way of doing computation but the 128 00:06:59,840 --> 00:07:04,477 benefit of course is that Everything isn't preconceived and pre-documented 129 00:07:04,560 --> 00:07:08,557 and it's it's contextually s contextually sensible and and therefore 130 00:07:08,640 --> 00:07:12,157 intelligent. >> So AI factories and robotics and we 131 00:07:12,240 --> 00:07:13,917 heard it yesterday from his royal 132 00:07:14,000 --> 00:07:16,877 highness his vision how to augment our 133 00:07:16,960 --> 00:07:19,677 workforce with roughly tens of millions 134 00:07:19,760 --> 00:07:22,637 of robotics to be able to infuse the 135 00:07:22,720 --> 00:07:25,277 next wave of productivity and progress. 136 00:07:25,360 --> 00:07:28,317 But this scares a lot of folks comes to 137 00:07:28,400 --> 00:07:32,877 the future of jobs. So let's hear about your thoughts uh Elon and Jensen on 138 00:07:32,960 --> 00:07:37,277 that. >> Uh sure. Well 139 00:07:37,360 --> 00:07:39,677 um if you say like in the long term 140 00:07:39,760 --> 00:07:44,317 where will things end up long term? I don't know what long term is. Maybe it's 141 00:07:44,400 --> 00:07:46,317 10 20 years something like that. For me 142 00:07:46,400 --> 00:07:51,437 that's long term. Um my prediction is that work will be optional. 143 00:07:51,520 --> 00:07:56,717 >> Optional. >> Optional. Um so 144 00:07:56,800 --> 00:08:00,717 >> we'll take that. >> Yeah. I mean it it'll be like uh playing 145 00:08:00,800 --> 00:08:02,397 sports or a video game or something like 146 00:08:02,480 --> 00:08:05,757 that. Um if you want to work uh you know 147 00:08:05,840 --> 00:08:07,997 in the same way like you can you can go 148 00:08:08,080 --> 00:08:12,877 to the store and just buy some vegetables or you could grow vegetables 149 00:08:12,960 --> 00:08:16,637 in your backyard. It's much harder to grow vegetables in your backyard but 150 00:08:16,720 --> 00:08:18,237 some people still do it because they 151 00:08:18,320 --> 00:08:20,797 like growing growing vegetables. Um that 152 00:08:20,880 --> 00:08:25,677 will be what work is like optional. Um now between now and then there's 153 00:08:25,760 --> 00:08:28,237 actually a lot of work to get to that point. Mhm. 154 00:08:28,320 --> 00:08:31,117 >> Um, and I always recommend people read 155 00:08:31,200 --> 00:08:34,077 read Yin Banks uh culture books to get a 156 00:08:34,160 --> 00:08:37,917 sense for what a probable 157 00:08:38,000 --> 00:08:41,437 positive AI future is like. Um, and 158 00:08:41,520 --> 00:08:43,757 interestingly in those books, money is 159 00:08:43,840 --> 00:08:47,037 no longer doesn't exist. 160 00:08:47,120 --> 00:08:51,677 It's kind of interesting. And I I my guess is in in if you go out long enough 161 00:08:51,760 --> 00:08:53,437 assuming there's a continued improvement 162 00:08:53,520 --> 00:08:57,277 in AI and robotics which seems likely 163 00:08:57,360 --> 00:09:00,317 the money will will stop being relevant 164 00:09:00,400 --> 00:09:03,357 at some point in the future. Um now 165 00:09:03,440 --> 00:09:06,237 there will still be constraints on power 166 00:09:06,320 --> 00:09:10,077 like electricity and mass. Uh the 167 00:09:10,160 --> 00:09:12,077 fundamental physics elements will still 168 00:09:12,160 --> 00:09:15,997 be still be constraints. Um but um I 169 00:09:16,080 --> 00:09:18,077 think at some point 170 00:09:18,160 --> 00:09:23,197 uh currency becomes irrelevant. 171 00:09:23,280 --> 00:09:25,757 >> Jensen, any thoughts? 172 00:09:25,840 --> 00:09:30,570 >> Um [laughter] 173 00:09:30,880 --> 00:09:34,957 by the way, the Nvidia earnings call is later today. [laughter] 174 00:09:35,040 --> 00:09:38,957 >> And by the way, since currency is relevant. 175 00:09:39,040 --> 00:09:43,197 >> Cheers. >> [laughter] 176 00:09:43,280 --> 00:09:49,240 >> Elon just wants to share with you some breaking [laughter] news. 177 00:09:51,470 --> 00:09:53,677 [laughter] The two of us who like to share some 178 00:09:53,760 --> 00:09:57,117 breaking news. Uh let's see. I I would 179 00:09:57,200 --> 00:10:01,117 say I would say there there's um different horizons you could look at. 180 00:10:01,200 --> 00:10:03,357 Everybody's jobs will be different. That 181 00:10:03,440 --> 00:10:08,717 I think that that's for sure. Uh how how the students learn will be different. um 182 00:10:08,800 --> 00:10:12,077 uh how people do their work will be different obviously because a lot of the 183 00:10:12,160 --> 00:10:14,557 things that that we do mundanely or 184 00:10:14,640 --> 00:10:17,357 arduously or very difficultly are going 185 00:10:17,440 --> 00:10:22,077 to be done very simply and and so we're going to be more productive from that s 186 00:10:22,160 --> 00:10:23,997 from that sense one of the things that I 187 00:10:24,080 --> 00:10:27,677 will say is that for most people or a 188 00:10:27,760 --> 00:10:31,837 company if some if your life becomes more productive 189 00:10:31,920 --> 00:10:34,317 and if the things that you're doing uh 190 00:10:34,400 --> 00:10:39,037 with great difficulty becomes simpler it is very likely because you have so many 191 00:10:39,120 --> 00:10:43,117 ideas, you'll have more time to go pursue things. It is my guess that Elon 192 00:10:43,200 --> 00:10:47,437 will be busier as a result of AI, I'm going to be busier as a result of AI. 193 00:10:47,520 --> 00:10:51,197 And the reason for that is because we have so many ideas we want to pursue, so 194 00:10:51,280 --> 00:10:55,037 many things that that we still have in our backlog inside our company that we 195 00:10:55,120 --> 00:10:59,037 can go pursue. If we were more productive, we can get to those things 196 00:10:59,120 --> 00:11:01,677 faster. And so in the near term I would 197 00:11:01,760 --> 00:11:06,077 say that that there is every evidence that that we will be more productive and 198 00:11:06,160 --> 00:11:10,477 yet still be busier because we have so many ideas. One thing that I will say 199 00:11:10,560 --> 00:11:13,037 give you give you some evidence is that 200 00:11:13,120 --> 00:11:14,477 uh and I was just telling Elon about 201 00:11:14,560 --> 00:11:17,677 this earlier radiology for example has 202 00:11:17,760 --> 00:11:20,797 largely been uh converted to AIdriven 203 00:11:20,880 --> 00:11:22,397 radiology and there's some really great 204 00:11:22,480 --> 00:11:25,517 companies doing that and the surprising 205 00:11:25,600 --> 00:11:28,477 thing is the prediction that all 206 00:11:28,560 --> 00:11:30,637 radiologists would be the first jobs to 207 00:11:30,720 --> 00:11:33,677 go was exactly the opposite. The trend 208 00:11:33,760 --> 00:11:39,117 shows that there are more radiologists being hired now as a result of AI. And 209 00:11:39,200 --> 00:11:42,877 the reason for that, if you take a step back, it's because the goal of a 210 00:11:42,960 --> 00:11:45,757 radiologist is not to study the images. 211 00:11:45,840 --> 00:11:48,237 The goal of a radiologist is to diagnose 212 00:11:48,320 --> 00:11:53,037 a disease. Now, the studying of the images became so productive. They could 213 00:11:53,120 --> 00:11:57,197 study more images, study more modalities, spend more time with the 214 00:11:57,280 --> 00:12:01,277 patients. And as a result, they were actually accepting more patients. We're 215 00:12:01,360 --> 00:12:04,237 doing more radiology all around the world. We're doing a better job with 216 00:12:04,320 --> 00:12:06,717 diagnosing disease. And so that's that's 217 00:12:06,800 --> 00:12:10,717 kind of the the nearterm near uh outcome 218 00:12:10,800 --> 00:12:13,917 of of uh AI and productivity. And and 219 00:12:14,000 --> 00:12:18,637 we'll see we'll see what happens long term. You know, I I 220 00:12:18,720 --> 00:12:22,397 when when currency doesn't matter anymore, just you know, let me know 221 00:12:22,480 --> 00:12:24,957 right before. 222 00:12:25,040 --> 00:12:29,437 >> You'll see it coming. [laughter] You'll see it coming like like 223 00:12:29,520 --> 00:12:31,677 >> we text often so just Yeah, we do. 224 00:12:31,760 --> 00:12:34,717 >> Yeah, just text it off. Yeah, let me know. I kind of I kind of agree with 225 00:12:34,800 --> 00:12:37,117 with both of you because if you look at 226 00:12:37,200 --> 00:12:40,237 every technological trend every general 227 00:12:40,320 --> 00:12:45,517 purpose technology has been net new positive for for the globe for humanity 228 00:12:45,600 --> 00:12:49,997 and so forth and let me share with you two things >> and your excellency I think it's 229 00:12:50,080 --> 00:12:54,397 precisely the reason the reason for that is because all the great ideas from from 230 00:12:54,480 --> 00:12:57,837 innovators like Elon you have so many good ideas 231 00:12:57,920 --> 00:13:01,677 >> that and Jason as well >> yeah well you know thank you 232 00:13:01,760 --> 00:13:05,517 >> so let me share with you two stories from two Saudi innovators in 233 00:13:05,600 --> 00:13:09,757 collaboration with a lot of great the great work that Nvidia does that gra 234 00:13:09,840 --> 00:13:12,157 does. So one of them is professor 235 00:13:12,240 --> 00:13:15,277 Omaragi who's the first American Saudi 236 00:13:15,360 --> 00:13:17,997 to win a Nobel prize in creating new 237 00:13:18,080 --> 00:13:20,477 chemistry and the way he has done that 238 00:13:20,560 --> 00:13:22,957 he has leveraged your AI accelerators 239 00:13:23,040 --> 00:13:27,597 and models like Grock to be able to create new chemistry when it comes to 240 00:13:27,680 --> 00:13:29,837 metal organic frameworks. Those are 241 00:13:29,920 --> 00:13:32,317 metal ions that are positively charged 242 00:13:32,400 --> 00:13:34,957 with organic linkers to be able to 243 00:13:35,040 --> 00:13:38,637 effectively create a sponge with 33 244 00:13:38,720 --> 00:13:42,077 nanometers pores to capture water from 245 00:13:42,160 --> 00:13:46,077 air and also to capture carbon dioxide. 246 00:13:46,160 --> 00:13:49,277 The second story has also to do with AI 247 00:13:49,360 --> 00:13:51,997 accelerated by Nvidia and with models 248 00:13:52,080 --> 00:13:54,637 like Brock which is nanop which is 249 00:13:54,720 --> 00:13:57,837 effectively creating a nano robot 500 250 00:13:57,920 --> 00:14:00,397 nanometers by a,000 nanometers to be 251 00:14:00,480 --> 00:14:03,197 able to do gene editing leveraging the 252 00:14:03,280 --> 00:14:05,757 crisper technology to take out sickle 253 00:14:05,840 --> 00:14:07,917 cell disease. Now in both these 254 00:14:08,000 --> 00:14:10,717 instances they originated 20 years ago 255 00:14:10,800 --> 00:14:14,477 in research but AI was able to really 256 00:14:14,560 --> 00:14:19,677 accelerate the outcomes and the outputs such that we can move into new value 257 00:14:19,760 --> 00:14:24,157 pools. So I think with every technological trend humanity is going to 258 00:14:24,240 --> 00:14:28,717 always manage to shift to new value pools when it comes to workforce and 259 00:14:28,800 --> 00:14:32,637 productivity. But we have some great announcements to talk about here today. 260 00:14:32,720 --> 00:14:36,557 Let's begin with you Elon. the things that we're doing with XAI. 261 00:14:36,640 --> 00:14:38,317 >> Uh yeah, we're excited to announce that 262 00:14:38,400 --> 00:14:43,437 um we're doing a a a uh a 500 megawatt I 263 00:14:43,520 --> 00:14:47,677 mean yeah 500 sorry >> 500 megawatt 264 00:14:47,760 --> 00:14:50,870 >> 500 megawatt. Yeah, >> we're doing [laughter] 265 00:14:51,200 --> 00:14:55,197 >> we're doing 500 megawatt. >> Sorry. Yeah. 266 00:14:55,280 --> 00:14:56,957 >> Yeah. Yeah. Yeah. The 500 g one will 267 00:14:57,040 --> 00:15:01,357 have to wait. Um so um that that'll be 268 00:15:01,440 --> 00:15:04,487 eight bazillion trillion dollars. Um, 269 00:15:04,570 --> 00:15:05,917 [laughter] 270 00:15:06,000 --> 00:15:11,830 stop that. >> Uh, [laughter] 271 00:15:11,920 --> 00:15:16,317 >> so yeah, we we're we're doing a um XAI 272 00:15:16,400 --> 00:15:19,117 and the King of Saudi Arabia are doing a a 273 00:15:19,200 --> 00:15:23,357 >> Humane 500 megawatt starting with 50 megawatt phase one and we're doing it 274 00:15:23,440 --> 00:15:25,757 with Nvidia. Congratulations to the 275 00:15:25,840 --> 00:15:28,157 Humane team to target team. Such a 276 00:15:28,240 --> 00:15:33,117 fantastic job. Jensen, I think we're also doing some great announcements. Uh 277 00:15:33,200 --> 00:15:37,437 this week >> we are [laughter] >> well we're we're announcing we're 278 00:15:37,520 --> 00:15:41,037 announcing all kinds of things. Um uh our partnership with Humane is is going 279 00:15:41,120 --> 00:15:44,077 incredibly well. First of all uh we we 280 00:15:44,160 --> 00:15:46,077 uh work together to get this company 281 00:15:46,160 --> 00:15:49,197 started and off the ground and just got 282 00:15:49,280 --> 00:15:52,237 an incredible customer with Elon. 283 00:15:52,320 --> 00:15:54,717 Could you imagine a startup company 284 00:15:54,800 --> 00:15:58,237 approximately 0 billion in revenues now 285 00:15:58,320 --> 00:16:03,197 going to uh build a data center for Elon? uh 500 megawatts is gigantic. Uh 286 00:16:03,280 --> 00:16:05,117 this company is off the charts uh right 287 00:16:05,200 --> 00:16:11,677 away. Uh in addition to that, we're um uh working working uh AWS as you know is 288 00:16:11,760 --> 00:16:16,237 also >> congratulations to the humane team with AWS 289 00:16:16,320 --> 00:16:18,157 starting with 100 megawatts with a 290 00:16:18,240 --> 00:16:21,437 gigawatt ambition and uh counting. 291 00:16:21,520 --> 00:16:23,997 >> So AWS is also coming to humane. Uh 292 00:16:24,080 --> 00:16:25,997 we're working with uh humane on 293 00:16:26,080 --> 00:16:29,117 omniverse uh digital twins. uh as you 294 00:16:29,200 --> 00:16:32,637 know that AI is not just well a just 295 00:16:32,720 --> 00:16:37,437 agentic AI and chat bots and uh cognitive AI is incredibly important to 296 00:16:37,520 --> 00:16:39,517 the world uh but AI applies to 297 00:16:39,600 --> 00:16:44,797 everything chemicals and proteins and genes and physics and fluid dynamics and 298 00:16:44,880 --> 00:16:48,797 particles and of course robotics and activation and um and we created this 299 00:16:48,880 --> 00:16:51,197 world called Omniverse where robots can 300 00:16:51,280 --> 00:16:54,317 learn how to be good robots and and it's 301 00:16:54,400 --> 00:16:58,957 physically based it obeys the laws of physics and so robots can learn in these 302 00:16:59,040 --> 00:17:00,317 environments and we're working with 303 00:17:00,400 --> 00:17:03,517 Humane to apply Omniverse to all kinds 304 00:17:03,600 --> 00:17:07,917 of uh digital factories and robotics and warehouses and things like that. And so 305 00:17:08,000 --> 00:17:12,397 that's that's another uh we're we're also working uh in Saudi Arabia to build 306 00:17:12,480 --> 00:17:14,637 supercomputers to simulate quantum 307 00:17:14,720 --> 00:17:17,837 computers and and uh uh using our 308 00:17:17,920 --> 00:17:22,237 computers to be the controller and the error correction quantum error 309 00:17:22,320 --> 00:17:26,317 correction requires an enormous amount of computation and so so we're doing a 310 00:17:26,400 --> 00:17:29,837 lot of great work there too. So, a big partnership with Humane. They're off the 311 00:17:29,920 --> 00:17:33,037 charts. Um, off the ground and off the charts at the same time. 312 00:17:33,120 --> 00:17:35,437 >> This is how we walk the talk in the 313 00:17:35,520 --> 00:17:37,197 Kingdom of Saudi Arabia in partnership 314 00:17:37,280 --> 00:17:40,477 with the US. Yesterday, the president 315 00:17:40,560 --> 00:17:42,477 and his Royal Highness announced the AI 316 00:17:42,560 --> 00:17:44,877 strategic framework and partnership. 317 00:17:44,960 --> 00:17:49,517 Today, we're going big with Elon and Jensen. So, thank you for those 318 00:17:49,600 --> 00:17:53,610 opportunities. [applause] 319 00:17:54,880 --> 00:17:57,357 Now they they told me I have time for 320 00:17:57,440 --> 00:18:00,237 two last questions. So last night at the 321 00:18:00,320 --> 00:18:02,717 dinner I got a number of questions 322 00:18:02,800 --> 00:18:07,277 because it seems that the schedule leaked and uh everybody was giving me 323 00:18:07,360 --> 00:18:09,037 hints about the last two questions I'm 324 00:18:09,120 --> 00:18:11,117 going to do. So the first one was for 325 00:18:11,200 --> 00:18:15,277 you Elon and there's a big one for you 326 00:18:15,360 --> 00:18:18,797 Jensen. So prepare for that one. AI in 327 00:18:18,880 --> 00:18:21,997 space is that possible? 328 00:18:22,080 --> 00:18:25,517 Uh yes, if if civilization continues, 329 00:18:25,600 --> 00:18:29,517 which it probably will, uh then AI in space is inevitable. 330 00:18:29,600 --> 00:18:32,557 Um 331 00:18:32,640 --> 00:18:37,517 um you know, I always have to like preface that, you know, 332 00:18:37,600 --> 00:18:41,517 we shouldn't take civilization for granted. We we need to make sure to take 333 00:18:41,600 --> 00:18:43,837 care to ensure that civilization has an 334 00:18:43,920 --> 00:18:48,557 an upward arc. I mean, any student of history knows that civilization does not 335 00:18:48,640 --> 00:18:53,277 always have an upward arc. And in fact, civilizations have lif life cycles. So 336 00:18:53,360 --> 00:18:55,357 hopefully we are in a strong upward arc. 337 00:18:55,440 --> 00:18:57,437 I think we are for now. Um but we don't 338 00:18:57,520 --> 00:18:58,717 want to take that for granted or be 339 00:18:58,800 --> 00:19:02,877 complacent. Um but the in order to the 340 00:19:02,960 --> 00:19:07,117 way to think of AI in space is that in order to achieve any meaningful 341 00:19:07,200 --> 00:19:09,037 percentage of a cadesev two scale 342 00:19:09,120 --> 00:19:11,757 civilization where you're using even a 343 00:19:11,840 --> 00:19:14,397 millionth a millionth of the sun's 344 00:19:14,480 --> 00:19:17,917 energy you must have solar powered AI 345 00:19:18,000 --> 00:19:22,317 satellites in in deep space. Um so so 346 00:19:22,400 --> 00:19:26,237 that once you realize like once you think in terms of a cottage two scale 347 00:19:26,320 --> 00:19:28,477 civilization which is what what 348 00:19:28,560 --> 00:19:30,877 percentage of the sun's energy are you 349 00:19:30,960 --> 00:19:33,757 turning into useful work um then you 350 00:19:33,840 --> 00:19:36,957 then it becomes obvious that space is 351 00:19:37,040 --> 00:19:39,517 overwhelmingly what matters 352 00:19:39,600 --> 00:19:41,757 overwhelmingly the the sun only receives 353 00:19:41,840 --> 00:19:44,957 one roughly 1/2 billionth of the earth 354 00:19:45,040 --> 00:19:48,077 only receives roughly u two billionth of 355 00:19:48,160 --> 00:19:50,077 the sun's energy 356 00:19:50,160 --> 00:19:53,197 So if you want to have something that is 357 00:19:53,280 --> 00:19:55,517 say a million times more energy than 358 00:19:55,600 --> 00:19:57,837 Earth could possibly produce, 359 00:19:57,920 --> 00:20:03,597 >> you must go into space. It's and and so 360 00:20:03,680 --> 00:20:07,837 um you know this is where it's kind of handy to have a space company, I guess. 361 00:20:07,920 --> 00:20:12,397 Um sell the book. >> Easier to cool chips in space, too. 362 00:20:12,480 --> 00:20:14,717 >> Yes. >> Easier to cool chips in space. Yeah. 363 00:20:14,800 --> 00:20:17,037 >> Yes. There's definitely no water in space, so you're going to have to do 364 00:20:17,120 --> 00:20:19,357 something. Yeah. >> Uh that doesn't involve water. 365 00:20:19,440 --> 00:20:21,837 >> Just hang out. >> Well, it's you just got to radiate. 366 00:20:21,920 --> 00:20:25,597 >> That's right. >> Um so 367 00:20:25,680 --> 00:20:27,277 my estimate is that actually that that 368 00:20:27,360 --> 00:20:32,717 that the cost of of electricity like 369 00:20:32,800 --> 00:20:35,677 like the cost effectiveness of AI in 370 00:20:35,760 --> 00:20:38,797 space will be overwhelmingly better than 371 00:20:38,880 --> 00:20:41,757 AI on the ground. So far long before you 372 00:20:41,840 --> 00:20:43,917 uh exhaust potential energy sources on 373 00:20:44,000 --> 00:20:46,557 on Earth. Long long before meaning like 374 00:20:46,640 --> 00:20:49,277 I think even perhaps in the 375 00:20:49,360 --> 00:20:52,237 four or five year time frame the lowest 376 00:20:52,320 --> 00:20:55,917 cost way to do AI compute will be with 377 00:20:56,000 --> 00:20:59,277 solar powered AI satellites. 378 00:20:59,360 --> 00:21:03,037 So I'd say not more than 5 years from now. 379 00:21:03,120 --> 00:21:06,317 >> Wow. >> And just look at the supercomputers 380 00:21:06,400 --> 00:21:10,157 we're building together. Let's say each one of the racks is two tons. Out of 381 00:21:10,240 --> 00:21:14,477 that two tons, 1.95 of it is probably for cooling, >> right? 382 00:21:14,560 --> 00:21:17,437 >> Oh, yeah. >> Just imagine how tiny that little 383 00:21:17,520 --> 00:21:21,117 supercomputer is, right? Each one of these GB300 racks will just be a little 384 00:21:21,200 --> 00:21:23,517 tiny thing. >> And and just electricity generation is 385 00:21:23,600 --> 00:21:25,437 is already becoming a challenge. 386 00:21:25,520 --> 00:21:28,957 >> Um so if if you start doing any kind of 387 00:21:29,040 --> 00:21:31,277 scaling for both electricity generation 388 00:21:31,360 --> 00:21:34,397 and cooling, um you realize, okay, space 389 00:21:34,480 --> 00:21:38,637 is incredibly compelling. Um so 390 00:21:38,720 --> 00:21:41,357 like let's say you wanted to do um I 391 00:21:41,440 --> 00:21:47,757 don't know two or 300 gawatt per year um of of AI compute. 392 00:21:47,840 --> 00:21:50,157 >> Yeah. >> Um it's very difficult to do that on 393 00:21:50,240 --> 00:21:53,357 Earth. Uh the so the the uh US average 394 00:21:53,440 --> 00:21:55,677 electricity usage uh last time I checked 395 00:21:55,760 --> 00:21:59,117 was around 460 gawatts per year average 396 00:21:59,200 --> 00:22:04,157 usage. Um so so something like say uh 397 00:22:04,240 --> 00:22:06,317 you know three 300 if you're doing 300 398 00:22:06,400 --> 00:22:11,437 gawatts a year that would be like 2/3 of US electricity production per year. 399 00:22:11,520 --> 00:22:15,517 >> There's no way you're building power plants at that level. Um and then if you 400 00:22:15,600 --> 00:22:18,797 take it up to say a terowatt per year impossible. Yeah. 401 00:22:18,880 --> 00:22:21,037 >> Like you have to do that in space. 402 00:22:21,120 --> 00:22:23,197 >> There there just is there just is no way 403 00:22:23,280 --> 00:22:26,637 to do a terowatt uh per year on Earth. 404 00:22:26,720 --> 00:22:28,157 Um 405 00:22:28,240 --> 00:22:30,797 and uh and and in space you you've got 406 00:22:30,880 --> 00:22:36,317 continuous solar. Um you've got uh you don't you actually don't need batteries 407 00:22:36,400 --> 00:22:37,757 because it's always sunny in space. 408 00:22:37,840 --> 00:22:41,597 >> Right. Exactly. >> Um and um and and the solar panels 409 00:22:41,680 --> 00:22:42,797 actually become cheaper because you 410 00:22:42,880 --> 00:22:48,797 don't need glass or framing. Um and the cooling is just radiative. So that's 411 00:22:48,880 --> 00:22:51,037 that's why I think >> that's the dream. 412 00:22:51,120 --> 00:22:53,997 >> Yes, >> that's the dream. So Jensen, everybody 413 00:22:54,080 --> 00:22:58,557 last night was asking me, and I'm mindful it's a earnings uh call for you 414 00:22:58,640 --> 00:23:02,477 today. So I'm going to say this delicately. Everybody has been asking me 415 00:23:02,560 --> 00:23:06,477 to ask you, are we going to have an AI bubble? [laughter] 416 00:23:06,560 --> 00:23:10,877 That's the last question. All right, let's No. 417 00:23:10,960 --> 00:23:14,717 All right, let me Well, let me just hear you what we see. Okay, so so I I think 418 00:23:14,800 --> 00:23:18,077 it's really important when you look at what's happening around the world and go 419 00:23:18,160 --> 00:23:21,037 back to first principles of what's happening in computer science and 420 00:23:21,120 --> 00:23:25,277 computing. There are three things that that's happening. The first thing is 421 00:23:25,360 --> 00:23:29,117 that we all know that Moore's law has run its course and the ability that the 422 00:23:29,200 --> 00:23:32,557 amount of demand for computing versus the amount of computation we can get out 423 00:23:32,640 --> 00:23:36,157 of general purpose computing is really challenging and so the world's been 424 00:23:36,240 --> 00:23:39,277 moving to accelerated computing for some time. We've been pushing this now for 425 00:23:39,360 --> 00:23:41,997 some over 20 years. Let me give you one 426 00:23:42,080 --> 00:23:44,637 statistic. I was just at supercomputing 427 00:23:44,720 --> 00:23:48,157 six years ago. Uh 428 00:23:48,240 --> 00:23:51,517 CPUs were 90% of the world's 429 00:23:51,600 --> 00:23:53,597 supercomputers, top 500 supercomputers 430 00:23:53,680 --> 00:23:58,157 six years ago. This year less than 15%. 431 00:23:58,240 --> 00:24:01,357 Went from 90% to 10%. And meanwhile, 432 00:24:01,440 --> 00:24:06,157 accelerated computing went from the other way, 10% to now 90%. Okay, so 433 00:24:06,240 --> 00:24:10,557 you're seeing that inflection point, the transition in high performance computing 434 00:24:10,640 --> 00:24:13,757 from general purpose computing to accelerated computing. Well, one of the 435 00:24:13,840 --> 00:24:17,437 one of the most data intensive one of the most intensive computation things 436 00:24:17,520 --> 00:24:21,437 that the world does in cloud is data processing. Several hundred billion 437 00:24:21,520 --> 00:24:23,997 dollars of computation is done on just 438 00:24:24,080 --> 00:24:27,037 raw data processing. That Rexus is the 439 00:24:27,120 --> 00:24:31,117 engine of the internet today that's going generative AI. It used to be 440 00:24:31,200 --> 00:24:33,357 running on CPUs now runs on GPUs. Which 441 00:24:33,440 --> 00:24:36,157 then says the third thing when if you 442 00:24:36,240 --> 00:24:39,197 just look at those two applications 443 00:24:39,280 --> 00:24:41,837 many of the internet companies can build 444 00:24:41,920 --> 00:24:44,477 enormous number of GPU supercomputers 445 00:24:44,560 --> 00:24:49,197 just doing that of course then it creates this the third opportunity on 446 00:24:49,280 --> 00:24:52,957 top of it which is agentic AI this is grock and this is open AI this is 447 00:24:53,040 --> 00:24:57,677 anthropic you know this is Gemini agentic AI sits on top of that but don't 448 00:24:57,760 --> 00:25:01,677 you know don't forget to think about what is happening 449 00:25:01,760 --> 00:25:04,637 above underneath ath what everybody sees 450 00:25:04,720 --> 00:25:09,197 as AI today. There's a whole movement of computing from general purpose computing 451 00:25:09,280 --> 00:25:12,797 to accelerated computing. And that if you just if you take that into 452 00:25:12,880 --> 00:25:14,557 consideration, you'll come to the 453 00:25:14,640 --> 00:25:17,277 conclusion that in fact what is left 454 00:25:17,360 --> 00:25:20,557 over to fuel that revolutionary agentic 455 00:25:20,640 --> 00:25:23,597 AI is not only substantially less than 456 00:25:23,680 --> 00:25:26,637 you thought and all of it justified. 457 00:25:26,720 --> 00:25:28,877 Well, I was just informed by the team 458 00:25:28,960 --> 00:25:31,677 that my boss and your bosses is going to 459 00:25:31,760 --> 00:25:36,397 talk next, the honorable president and his royal highness the crown prince and 460 00:25:36,480 --> 00:25:38,637 hence we ran out of time. But in 461 00:25:38,720 --> 00:25:41,837 essence, [applause] 462 00:25:41,920 --> 00:25:45,997 this is a such so much love for you Elon 463 00:25:46,080 --> 00:25:49,197 and and Jensen. But this in essence is a 464 00:25:49,280 --> 00:25:51,677 92 alliance 465 00:25:51,760 --> 00:25:54,797 that shifted from energy to digital to 466 00:25:54,880 --> 00:25:57,677 the intelligence age powered by pioneers 467 00:25:57,760 --> 00:26:00,157 such as Elon and Jensen to serve 468 00:26:00,240 --> 00:26:03,197 humanity and create on a net new basis 469 00:26:03,280 --> 00:26:05,757 new economies, new jobs and a better 470 00:26:05,840 --> 00:26:10,317 future for humanity powered by the Kingdom of Saudi Arabia and the United 471 00:26:10,400 --> 00:26:14,077 States. Thank you for our lifetime partnership and friendship. Thank you 472 00:26:14,160 --> 00:26:17,177 Elon. Thank you Jensen. >> Thank you. 473 00:26:17,260 --> 00:26:19,280 >> [applause] 474 00:26:20,530 --> 00:26:22,550 [music]