1 00:00:00,160 --> 00:00:04,960 Speaker 1: Latest interview of Elon Musk, like, how do you decide 2 00:00:06,759 --> 00:00:08,679 what progress the civilization has made? 3 00:00:09,560 --> 00:00:16,199 Speaker 2: That's the most objective metric that any alien species, say 4 00:00:16,280 --> 00:00:21,280 visiting us, would calibrate how much progress. 5 00:00:20,920 --> 00:00:23,000 Speaker 1: We've made as a civilization. 6 00:00:23,600 --> 00:00:25,239 Speaker 2: And one of the most objective ways to do that 7 00:00:25,440 --> 00:00:30,920 is the amount of power that is any given civilization 8 00:00:31,079 --> 00:00:36,240 has been able to harness. And there was a Russian 9 00:00:36,320 --> 00:00:40,119 physicist actually I think by the name of Kadashov, who 10 00:00:40,200 --> 00:00:43,079 thought about this, and it's I think it's a good 11 00:00:43,119 --> 00:00:44,359 way to characterize it, which is. 12 00:00:47,159 --> 00:00:47,960 Speaker 1: You can have. 13 00:00:49,439 --> 00:00:54,159 Speaker 2: You can assess how well a civilization is harnessing the 14 00:00:54,200 --> 00:00:58,439 power available on the planet. That's a type one, and 15 00:00:58,479 --> 00:01:02,679 then type two would be how much of the stars 16 00:01:03,200 --> 00:01:05,640 power are you honessing? And then type three would be 17 00:01:05,719 --> 00:01:10,040 how much of the galaxy's power are you honessing? These 18 00:01:10,040 --> 00:01:15,959 are very objective and measurable numbers. So right now we're 19 00:01:16,480 --> 00:01:19,480 very low on Na Kadashev one scale, and if you say, 20 00:01:19,519 --> 00:01:25,439 like what proportion of our planet's power are we honessing, 21 00:01:25,799 --> 00:01:31,840 it's a very very tiny number. And basically we're harnessing 22 00:01:32,159 --> 00:01:38,200 almost nothing of our stars power. So the sun is 23 00:01:38,280 --> 00:01:43,000 truly an immense thing. It is it's difficult with words 24 00:01:43,040 --> 00:01:46,040 to characterize just how immense the Sun is, but this 25 00:01:46,120 --> 00:01:47,760 gives you sort of a sense of scale. 26 00:01:48,000 --> 00:01:52,000 Speaker 3: Yeah, it's a big difficulty jump going from level one 27 00:01:52,000 --> 00:01:52,840 to level two. 28 00:01:53,400 --> 00:01:54,599 Speaker 1: Very big difficulty jump. 29 00:01:54,719 --> 00:01:56,480 Speaker 2: Yes, and level three and we don't even know how 30 00:01:56,519 --> 00:01:58,159 to do level three really well. 31 00:01:58,200 --> 00:02:01,159 Speaker 1: Get yeah, yeah, exactly a. I'll figure it out. Yeah. 32 00:02:01,400 --> 00:02:03,959 Speaker 2: One way to appreciate the size of the Sun is 33 00:02:04,000 --> 00:02:06,719 to think about how heavy is the Sun compared to 34 00:02:06,799 --> 00:02:09,439 all the rest of the mass in the Solar System. 35 00:02:09,759 --> 00:02:12,280 So the Sun is about ninety nine point eighty six 36 00:02:12,319 --> 00:02:17,639 percent of all mass in the Solar system, it's everything, 37 00:02:17,800 --> 00:02:21,759 and then off the remaining one zero point one four percent, 38 00:02:22,520 --> 00:02:26,159 most of that is Jupiter, one planet, So we're. 39 00:02:26,000 --> 00:02:26,719 Speaker 3: A sole lightweight. 40 00:02:28,080 --> 00:02:31,759 Speaker 2: Yes, the entire mass of Earth is in the tiny 41 00:02:31,800 --> 00:02:35,159 miss laneous category, where we're like, Earth is a tiny 42 00:02:35,240 --> 00:02:37,879 dust mote compared to the Sun. 43 00:02:38,159 --> 00:02:41,840 Speaker 3: Well, but how much energy are we talking like coming 44 00:02:41,879 --> 00:02:46,000 from the Sun, especially compared to what we're using here 45 00:02:46,039 --> 00:02:46,479 on Earth. 46 00:02:46,919 --> 00:02:50,639 Speaker 2: It feels like, Yeah, the incident's solar energy on the 47 00:02:50,680 --> 00:02:53,120 cross section of the Earth is roughly a half billionth 48 00:02:53,319 --> 00:02:59,120 of the Sun's power output, and the vast majority of 49 00:02:59,159 --> 00:03:02,680 that we cannot because you know, seventy percent of Earth 50 00:03:02,759 --> 00:03:06,039 is water. We should technically our plan should be called 51 00:03:06,120 --> 00:03:09,639 water because that is seventy percent water. And I think 52 00:03:09,680 --> 00:03:12,280 an alien civilization visiting us would be like, why are 53 00:03:12,280 --> 00:03:14,080 they're calling it Earth when it is mostly water? 54 00:03:14,599 --> 00:03:17,599 Speaker 3: Where the greenland's not green? Of the of the gally 55 00:03:17,599 --> 00:03:19,479 of the solar system. 56 00:03:19,080 --> 00:03:25,039 Speaker 2: Yeah, a bunch of the exactly even, we're seventy percent water, 57 00:03:25,599 --> 00:03:27,840 and then of the thirty percent best land, a bunch 58 00:03:27,879 --> 00:03:33,039 of it is either Antarctica or you know, Siberia type 59 00:03:33,080 --> 00:03:35,759 of thing, very northern Canada type of thing, very difficult 60 00:03:35,759 --> 00:03:38,439 to not places people typically. 61 00:03:38,039 --> 00:03:38,479 Speaker 1: Want to live. 62 00:03:38,840 --> 00:03:40,919 Speaker 2: And you're not going to get a lot of solar 63 00:03:40,960 --> 00:03:45,159 power at the polls. So the actual usable area of 64 00:03:46,159 --> 00:03:50,199 land that where you can get solar power is quite small. Anyway, 65 00:03:50,360 --> 00:03:54,879 in order to ascend the Cordship scale, in order to 66 00:03:55,120 --> 00:03:59,479 get to any meaningful percentage of the Sun's energy harnessed, 67 00:04:00,280 --> 00:04:02,520 you have to go to space. If you wanted to 68 00:04:02,520 --> 00:04:06,719 get to say, a millionth of the power output of 69 00:04:06,719 --> 00:04:12,560 the Sun, you would have to increase civilizational energy harnessed 70 00:04:12,639 --> 00:04:14,919 by much more than a million. 71 00:04:16,519 --> 00:04:20,079 Speaker 1: So we currently use much less than a trillionth. 72 00:04:20,240 --> 00:04:23,839 Speaker 2: Of the power output of the Sun, and a trillion 73 00:04:23,920 --> 00:04:27,800 is a million times a million. So so basically this 74 00:04:28,800 --> 00:04:33,120 we're basically practically nowhere on the sort of the Cottagship 75 00:04:33,199 --> 00:04:34,600 to scale, practically nowhere. 76 00:04:34,759 --> 00:04:37,160 Speaker 3: So in CARDISHIV scale we're all still. 77 00:04:39,959 --> 00:04:44,879 Speaker 2: Register We're like not, we're not even yeah, we're so 78 00:04:45,639 --> 00:04:49,279 we're not We're not registering not even a microsol. 79 00:04:49,600 --> 00:04:50,639 Speaker 1: Yeah no. 80 00:04:51,720 --> 00:04:54,439 Speaker 2: And so to actually what microsol would be an epic, 81 00:04:54,560 --> 00:04:56,000 epic achievement relative. 82 00:04:55,759 --> 00:04:58,040 Speaker 3: To where we are right now, something to aspire to. 83 00:04:58,399 --> 00:04:59,920 Speaker 1: Yeah, yeah, that's o gold. 84 00:05:00,639 --> 00:05:07,040 Speaker 2: And like this is I think both simultaneously an incredibly 85 00:05:07,120 --> 00:05:11,680 adventurous goal relatives where we are and yet not particularly 86 00:05:11,680 --> 00:05:14,360 adventurous as a percentage of the Sun's energy to. 87 00:05:14,319 --> 00:05:15,480 Speaker 1: Try to achieve. 88 00:05:16,639 --> 00:05:19,759 Speaker 2: Powerhunness being one million of what the Sun outputs. 89 00:05:20,399 --> 00:05:23,279 Speaker 1: And so to actually start a microstal that actually. 90 00:05:23,079 --> 00:05:25,600 Speaker 3: Start getting there though we're not just gonna throw a 91 00:05:25,600 --> 00:05:27,800 solar raised in space try to soak up a bunch 92 00:05:27,839 --> 00:05:30,120 of the sun. Like there has to be a need, 93 00:05:30,360 --> 00:05:33,000 like you want to go up there and do something meaningful, 94 00:05:33,360 --> 00:05:36,560 And obviously until this point in human history, like there 95 00:05:36,600 --> 00:05:41,279 hasn't really been a need. What has changed to make 96 00:05:41,360 --> 00:05:43,839 us think that like, maybe now's the time to start 97 00:05:44,000 --> 00:05:46,680 trying to notch a percentage point or two, I mean. 98 00:05:46,680 --> 00:05:50,800 Speaker 2: Getting too a percent of the Sun's energy, maybe not 99 00:05:50,879 --> 00:05:53,319 a percentity Let's go like, well, well the desci will 100 00:05:53,319 --> 00:05:56,279 point back. So you're extremely thick ass civilization if you 101 00:05:56,319 --> 00:05:58,399 get one percent of the Sun's energy. And I'm like, wow, 102 00:05:59,199 --> 00:06:04,519 that civilizations be vastly more powerful in us, to say 103 00:06:04,560 --> 00:06:09,519 the least. So in order to start to make some 104 00:06:09,560 --> 00:06:14,000 progress on the Cottashef scale, we need to launch satellites 105 00:06:14,040 --> 00:06:19,759 to to to orbit Earth and capture solar power. And 106 00:06:20,839 --> 00:06:24,240 that avoids the need to build massive power plants on 107 00:06:24,319 --> 00:06:28,959 Earth and deal with cooling because cooling is actually much 108 00:06:29,000 --> 00:06:32,319 easier in space than it is on Earth. You can 109 00:06:32,399 --> 00:06:38,399 just radiate to the vacuum. And and so what what 110 00:06:38,560 --> 00:06:40,639 we're proposing here and what we intend to do, is 111 00:06:41,160 --> 00:06:44,360 to try to claim the Cottagechef scale to be kind 112 00:06:44,360 --> 00:06:49,639 of like a respectable civilization. So when the aliens, hopefully 113 00:06:49,639 --> 00:06:53,399 there are aliens out there, and they maybe finally decide 114 00:06:53,439 --> 00:06:56,439 to talk to us, you know where we have where 115 00:06:56,480 --> 00:07:00,000 where we have some respectable amount of the Sun's energy 116 00:07:00,040 --> 00:07:05,680 being used, that's not like totally pathetic, which is the 117 00:07:05,720 --> 00:07:06,439 current situation. 118 00:07:06,879 --> 00:07:10,519 Speaker 3: And so before we start sending data centers, sending all 119 00:07:10,560 --> 00:07:13,439 of this to space, there are some limiting factors that 120 00:07:13,519 --> 00:07:16,480 we got to get there that would traditionally make it 121 00:07:16,519 --> 00:07:18,680 so like this is almost as possible. 122 00:07:18,759 --> 00:07:22,000 Speaker 2: Yeah, what does it take to scale? Yeah? So things 123 00:07:22,000 --> 00:07:25,279 it takes to scale are you need to have a 124 00:07:25,360 --> 00:07:28,160 large mask to orbit capability, which is what Starship will 125 00:07:28,199 --> 00:07:28,560 give us. 126 00:07:29,319 --> 00:07:31,519 Speaker 1: That large mask. So you know, you. 127 00:07:31,560 --> 00:07:35,439 Speaker 2: Ultimately need to send millions of tons to orbit and beyond, 128 00:07:35,839 --> 00:07:39,560 and you need the power associated with that. So if 129 00:07:39,560 --> 00:07:41,839 you want to put one hundred big wettes or ultimately 130 00:07:41,879 --> 00:07:45,600 a tarwowat into space from Earth, you need you will 131 00:07:45,600 --> 00:07:49,319 at some point need a tarowatt of solar and then 132 00:07:49,360 --> 00:07:51,639 you're going to need a tirawot of AI chips. So 133 00:07:51,680 --> 00:07:53,759 the three things you need a mask to orbit, a 134 00:07:53,800 --> 00:07:59,279 lot of solar power and radiators of course, and a 135 00:07:59,279 --> 00:07:59,879 lot of chips. 136 00:08:00,560 --> 00:08:03,000 Speaker 3: All right, well let's start ticking down the list. So 137 00:08:03,879 --> 00:08:07,800 mass orbit that's where Starship comes in. We just had 138 00:08:08,439 --> 00:08:12,680 first flight V three. It's awesome. I know you were there. 139 00:08:13,160 --> 00:08:16,199 It was crazy to see that rocket launch and like, 140 00:08:16,360 --> 00:08:19,160 long time coming. What's kind of what Starship's kind of 141 00:08:19,240 --> 00:08:21,759 purpose of being, What is it going to be doing? 142 00:08:23,240 --> 00:08:29,399 Speaker 2: Yeah, so Starship is going to it's going to revolutionize space. Really, 143 00:08:29,439 --> 00:08:34,039 it's the first rocket design that is capable of full 144 00:08:34,080 --> 00:08:38,639 and rapid reusability. Now, reusability is the fundamental breakthrough that 145 00:08:38,759 --> 00:08:42,120 is necessary to make life multiplanetary as well as to 146 00:08:42,200 --> 00:08:45,279 ascend the Cottership scale. You simply cannot ascend the Cottership 147 00:08:45,320 --> 00:08:49,720 scale unless you have a reusable spacecraft, and you cannot 148 00:08:50,120 --> 00:08:53,159 extend life to the Moon, to Mars and rest the 149 00:08:53,200 --> 00:08:55,159 Solar System without a reusable rocket. 150 00:08:56,679 --> 00:08:58,120 Speaker 1: The cost is simply prohibitive. 151 00:08:58,600 --> 00:09:01,519 Speaker 2: You can't You can't make enough rockets unless you fly, 152 00:09:01,720 --> 00:09:04,320 unless you can refly them. Just like any other mode 153 00:09:04,360 --> 00:09:07,200 of transport. You can imagine that if we had to 154 00:09:07,200 --> 00:09:10,440 throw away airplanes every time we flew, flying would be 155 00:09:10,519 --> 00:09:13,360 far too expensive and basically no one would be flying airplanes. 156 00:09:13,399 --> 00:09:14,600 Speaker 3: You're doing a whole lot more driving. 157 00:09:16,480 --> 00:09:21,919 Speaker 2: Yes, every mode of transport is reusable with that which 158 00:09:22,000 --> 00:09:31,080 is simply not viable as a transport system. So cars, planes, boats, forces, bicycles. 159 00:09:30,720 --> 00:09:32,080 Speaker 1: Are all obviously reusable. 160 00:09:33,399 --> 00:09:36,320 Speaker 2: With rockets, it's much harder to make a rocket reusable 161 00:09:36,360 --> 00:09:40,879 because Earth has a deep gravity well and a thick atmosphere, 162 00:09:41,559 --> 00:09:45,240 and these make it just barely possible to achieve reusability 163 00:09:45,279 --> 00:09:49,639 with a rocket. And there have been many prior attempts 164 00:09:49,720 --> 00:09:55,120 to create a fully reusable rocket, and they most of 165 00:09:55,120 --> 00:09:57,919 those attempts have been abandoned part way through because they 166 00:09:58,320 --> 00:10:02,679 didn't think they could succeed. In order to achieve full reusability, 167 00:10:02,879 --> 00:10:06,200 everything has got to be perfect, the engines, the structure, 168 00:10:06,519 --> 00:10:13,000 the avionics, the choice of propellant. You've got to you've 169 00:10:13,039 --> 00:10:15,759 got to go to extreme measures for mass optimization, which 170 00:10:15,759 --> 00:10:18,000 is why we have the tower catch the rocket instead 171 00:10:18,000 --> 00:10:19,320 of putting on landing. 172 00:10:19,039 --> 00:10:21,840 Speaker 1: Legs which are heavy. The rocket can s if we 173 00:10:21,879 --> 00:10:22,799 be caught by the tower. 174 00:10:23,080 --> 00:10:26,000 Speaker 2: And we haven't achieved full reusability yet, but we do 175 00:10:26,120 --> 00:10:29,519 expect to achieve that hopefully later this year with Starship. 176 00:10:29,759 --> 00:10:32,360 Speaker 1: And then you've got to achieve full reusability. They've also 177 00:10:32,919 --> 00:10:33,919 if you got to go step. 178 00:10:33,720 --> 00:10:38,039 Speaker 2: Beyond that, which is make it rapidly reuseable, such that 179 00:10:38,799 --> 00:10:41,720 the rocket lands, gets caught by the tower, is put 180 00:10:42,120 --> 00:10:44,480 back on the launch stand and can be flown again 181 00:10:44,720 --> 00:10:48,360 without any refurbishment or laborious. 182 00:10:47,799 --> 00:10:51,840 Speaker 1: Inspection like an aircraft. Yeah, this is incredibly difficult. 183 00:10:52,360 --> 00:10:54,159 Speaker 2: This is the first time that there's ever been a 184 00:10:54,240 --> 00:10:56,200 rocket where that is possible. 185 00:10:57,000 --> 00:10:58,799 Speaker 1: That's what makes starships so profound. 186 00:10:59,000 --> 00:11:03,360 Speaker 2: It also happen is to be the largest flying object 187 00:11:03,399 --> 00:11:06,840 ever made, the heaviest playing object ever made, the most 188 00:11:06,879 --> 00:11:10,759 powerful moving object of any kind. Starship V three is 189 00:11:12,000 --> 00:11:14,000 more than double the thrust of it the Saturn five 190 00:11:14,039 --> 00:11:18,080 Moon rocket. By version four will be pretty much three 191 00:11:18,080 --> 00:11:20,720 times the thrust of a Saturn five Moon rocket, and 192 00:11:20,759 --> 00:11:24,600 we expect this, We expect Starship to be flying more 193 00:11:24,600 --> 00:11:26,200 than once per hour down the road. 194 00:11:26,440 --> 00:11:28,440 Speaker 3: One of the fun facts from flight twelve that was 195 00:11:28,759 --> 00:11:32,919 actually the heaviest payload SpaceX has ever flown, and that's 196 00:11:32,919 --> 00:11:36,120 still just a fraction of what V three can do. 197 00:11:36,399 --> 00:11:43,240 So yes, I mean once we're flying massive amounts really rapidly. 198 00:11:43,360 --> 00:11:46,240 I mean we already fly the majority of payload to 199 00:11:46,240 --> 00:11:53,320 space with Falcon Do people even really understand what mass 200 00:11:53,480 --> 00:11:56,200 or of it becomes one starship is flying. 201 00:11:57,559 --> 00:12:00,679 Speaker 2: It's many orders ninety two greater than and what is 202 00:12:00,919 --> 00:12:05,279 the case today. So even with Falcon nine, Falcon Heavy, 203 00:12:05,960 --> 00:12:10,919 SpaceX levers almost ninety percent of all Earth mass to orbit. 204 00:12:11,759 --> 00:12:15,559 I think between eighty five ninety percent right now, and 205 00:12:15,600 --> 00:12:18,240 then most of the remaining mass I think is launched 206 00:12:18,240 --> 00:12:20,240 by China, and then the rest of the world, including 207 00:12:20,240 --> 00:12:23,279 the rest of the US, is the remaining I don't know, 208 00:12:23,440 --> 00:12:30,000 five to seven percent. Now with with Starship, we'll be 209 00:12:30,039 --> 00:12:33,240 aiming to go from somewhere around twenty five hundred tons 210 00:12:33,360 --> 00:12:36,200 a year to orbit to millions of tons per year 211 00:12:36,240 --> 00:12:40,120 to orbit, and to do so at a pretty short 212 00:12:40,240 --> 00:12:43,639 period of time. So we think probably we can get 213 00:12:43,679 --> 00:12:51,000 to a million tons to RBOT per year in about 214 00:12:51,360 --> 00:12:54,200 three years thereabouts Starship. 215 00:12:54,799 --> 00:12:57,240 Speaker 3: Starship is going to take care of the master or 216 00:12:57,240 --> 00:13:01,200 a bit limiting factor, yes, and then power generation, So 217 00:13:01,440 --> 00:13:05,919 first and A and maybe you can help people probably 218 00:13:05,960 --> 00:13:08,080 struggle to visualize a little bit when you say, like 219 00:13:08,159 --> 00:13:11,039 data center in space, Like we're not going to slap 220 00:13:11,320 --> 00:13:13,399 engines on a building and fly it up there, Like 221 00:13:13,440 --> 00:13:16,879 these actually look like pretty different and so kind of 222 00:13:16,919 --> 00:13:20,200 walk through how you take something that's in a giant 223 00:13:20,200 --> 00:13:22,840 building on the ground and turn it into something that's 224 00:13:22,879 --> 00:13:23,799 functional in space. 225 00:13:24,360 --> 00:13:26,240 Speaker 1: Yeah, I think it's it's pretty interesting. 226 00:13:26,279 --> 00:13:28,159 Speaker 4: A lot of people don't actually know what the inside 227 00:13:28,159 --> 00:13:29,000 of a data center. 228 00:13:28,799 --> 00:13:32,679 Speaker 2: Even looks like, right, and it's a mythical place where 229 00:13:33,200 --> 00:13:34,480 the Internet's in the cloud. 230 00:13:34,240 --> 00:13:36,919 Speaker 4: Or yeah, some people in vision wire, some people in 231 00:13:37,039 --> 00:13:41,000 vision boxes. But like, effectually it comes down to a 232 00:13:41,039 --> 00:13:44,279 set number of chips and the things that we need 233 00:13:44,320 --> 00:13:47,000 to launch into space are actually quite small when we 234 00:13:47,039 --> 00:13:49,960 look at it. The more challenging part is figuring out 235 00:13:49,960 --> 00:13:51,919 how to get how do you get the power for it? 236 00:13:52,320 --> 00:13:54,919 And that's where a lot of what we've worked on 237 00:13:55,159 --> 00:14:00,919 for existing like star wink technology, the solar rays are 238 00:14:00,960 --> 00:14:04,039 what we want to utilize that expertise to to be 239 00:14:04,080 --> 00:14:06,279 able to build a satellite that can actually launch the 240 00:14:06,320 --> 00:14:11,600 critical components of the data center into space itself. We 241 00:14:11,759 --> 00:14:14,279 like to look at this and say, like what is 242 00:14:14,320 --> 00:14:17,039 what is the actual engineering problem here? And and it's 243 00:14:17,120 --> 00:14:20,919 it's really a combination of delivering power and then taking 244 00:14:20,960 --> 00:14:23,279 the waste heat and energy away and sending it into 245 00:14:23,279 --> 00:14:27,159 the vacuum of space. As you mentioned, Yeah, uh, Now, 246 00:14:27,360 --> 00:14:32,279 the the AI satellite is actually much simpler than a 247 00:14:32,360 --> 00:14:36,360 starling satellite. It's a stalling satellite, has has gigantic phase 248 00:14:36,480 --> 00:14:41,600 ray antennas. Uh, it's got uh you know, parabolic antennas. 249 00:14:41,639 --> 00:14:46,159 It's got uh, you know a lot of laser links. 250 00:14:47,879 --> 00:14:51,120 It's a it's it's much more complicated than an AI satellite, 251 00:14:51,120 --> 00:14:54,639 and AI satellite is essentially a lot of solar cells, 252 00:14:55,519 --> 00:14:59,200 a radiator, and you still need some laser links, but 253 00:14:59,240 --> 00:15:03,559 you don't have of the super complex antennas. 254 00:15:03,080 --> 00:15:04,480 Speaker 1: That you have on a stalling satellite. 255 00:15:04,679 --> 00:15:07,840 Speaker 2: So I mean, given the two, the easier one to 256 00:15:08,000 --> 00:15:10,799 design for is the the AI satellite. 257 00:15:11,080 --> 00:15:13,919 Speaker 1: Yeah, it's just a little bit bigger. It's bigger. 258 00:15:14,240 --> 00:15:15,120 Speaker 3: Just makes stuff bigger. 259 00:15:15,200 --> 00:15:15,399 Speaker 1: Yeah. 260 00:15:15,440 --> 00:15:19,360 Speaker 3: I was like, so we've got this is our AI 261 00:15:19,480 --> 00:15:20,519 one if you guys want to. 262 00:15:20,480 --> 00:15:22,320 Speaker 1: Walk us through. Yeah. 263 00:15:22,360 --> 00:15:24,840 Speaker 4: So the first thing that we're really looking at here 264 00:15:24,960 --> 00:15:27,759 is like, first, you've got to make something compelling, right, 265 00:15:27,840 --> 00:15:30,440 And we thought that the right place to start is 266 00:15:31,120 --> 00:15:32,399 around one hundred and fifty. 267 00:15:32,200 --> 00:15:34,320 Speaker 1: Kilowatt like peak power level. 268 00:15:35,080 --> 00:15:37,440 Speaker 4: But as we look at the workloads with our experience 269 00:15:37,440 --> 00:15:41,759 with XAI, we get to actually see that we can 270 00:15:41,799 --> 00:15:44,519 also support about one hundred and twenty kilowats of average compute. 271 00:15:44,559 --> 00:15:47,279 Speaker 1: There's a difference. What we're showing here is kind of 272 00:15:47,360 --> 00:15:52,279 a draft version of the. 273 00:15:51,120 --> 00:15:54,480 Speaker 2: Version one of the of the SpaceX AI satellite AI one, 274 00:15:54,519 --> 00:15:57,600 I guess you could call it, and seems like a 275 00:15:57,600 --> 00:16:01,440 reasonable place to start is undred and fifty kilowatts peak 276 00:16:01,519 --> 00:16:05,039 power one hundred and twenty kilowatt sustained power. And to 277 00:16:05,080 --> 00:16:06,600 give you a sense of what does that actually look 278 00:16:06,679 --> 00:16:08,440 like in terms of the size of the. 279 00:16:08,480 --> 00:16:12,240 Speaker 1: Radiator size of the solar panels, the assumptions. 280 00:16:11,759 --> 00:16:14,919 Speaker 2: Here are two hundred and fifty what's per square meter 281 00:16:15,080 --> 00:16:19,440 for the solar array and about fourteen hundred what's per 282 00:16:19,519 --> 00:16:22,000 square meter for the radiators. So the radiators, these are 283 00:16:22,000 --> 00:16:27,399 double side radiators, are radiating both sides, they're oriented knife 284 00:16:27,480 --> 00:16:32,480 edge to the Sun, and it's fourteen hundred washpos square 285 00:16:32,480 --> 00:16:35,919 meter is a very achievable goal. Over time, we think 286 00:16:35,919 --> 00:16:38,279 we can probably do above chatter fifty washed post square 287 00:16:38,320 --> 00:16:41,639 meter and above fourteen hundred what's per square meter for 288 00:16:41,679 --> 00:16:44,840 the solar panels radiators, respectively. 289 00:16:45,919 --> 00:16:49,519 Speaker 1: But this gives you like a this is pretty much 290 00:16:49,519 --> 00:16:51,840 what the satellite's going to look like. Yeah, it's a lot. 291 00:16:51,799 --> 00:16:55,279 Speaker 2: Of solar panels, radiator, and then everything else is pretty small. 292 00:16:55,120 --> 00:16:55,720 Speaker 1: Like embarrassy. 293 00:16:56,000 --> 00:16:58,799 Speaker 4: And these are like evolutions of things that we have 294 00:16:58,919 --> 00:17:03,600 actually already launched in our Starlink constellation to date. Yeah, 295 00:17:03,879 --> 00:17:05,880 that's that's really I think the cool part to me 296 00:17:06,000 --> 00:17:09,319 is that we're looking at solar technology that we already 297 00:17:09,640 --> 00:17:13,319 are going to use on the V three Starlink vehicle. 298 00:17:13,440 --> 00:17:17,279 So I'm like really excited to then just take those 299 00:17:17,279 --> 00:17:18,079 and make it bigger. 300 00:17:18,960 --> 00:17:21,279 Speaker 1: Yeah. Part of what we want to convey here is 301 00:17:21,319 --> 00:17:22,400 that there's not some. 302 00:17:24,119 --> 00:17:28,359 Speaker 2: Magic that's necessary that doesn't exist for the AI satellites. 303 00:17:29,519 --> 00:17:31,000 As I had said, this is a lot of this 304 00:17:31,160 --> 00:17:35,839 is technology we've already made for the Starlink B three satellites. 305 00:17:36,920 --> 00:17:40,680 So it's basically, don't think this is a super hard 306 00:17:40,720 --> 00:17:42,599 problem compared to things we already do. 307 00:17:44,079 --> 00:17:45,599 Speaker 1: They would also be probably something on the. 308 00:17:45,640 --> 00:17:49,759 Speaker 2: Order of a twer bit of connectivity of laser link connectivity. 309 00:17:49,160 --> 00:17:51,319 Speaker 1: From the on the satellite. 310 00:17:52,480 --> 00:17:56,559 Speaker 2: One hundred and fifty killawak peak power level is roughly 311 00:17:56,599 --> 00:17:57,559 matches Le's ay. 312 00:17:57,680 --> 00:18:02,000 Speaker 1: An Nvidio GV three hundred rack would do, so a. 313 00:18:01,960 --> 00:18:05,720 Speaker 2: GB three hundred with seventy two GPUs it's peak power 314 00:18:05,720 --> 00:18:08,200 I think is around one hundred and forty kilowatts, but 315 00:18:08,240 --> 00:18:10,480 it's rarely it's it's almost impossible to get it to 316 00:18:11,079 --> 00:18:15,519 be at that pep power. A more reasonable operating envelope 317 00:18:15,559 --> 00:18:18,279 would be around one hundred and twenty twenty kilos average power, 318 00:18:19,480 --> 00:18:21,839 but it can peak up to one hundred and fifty. 319 00:18:21,920 --> 00:18:25,480 So that's it's basically thinking about as a rack of 320 00:18:25,480 --> 00:18:30,319 compute in space. And then you can connect these these 321 00:18:30,400 --> 00:18:34,160 racks of compute to either each other by the laser 322 00:18:34,160 --> 00:18:38,720 links or directly to the stalling constellations. So you can 323 00:18:38,759 --> 00:18:42,640 close the link with the stalling constellation, and then Stalling 324 00:18:42,880 --> 00:18:47,519 can then send that data to the ground using the 325 00:18:47,559 --> 00:18:51,920 existing KA and KU antennas on the on the vehicle, 326 00:18:52,720 --> 00:18:56,039 it also has laser links to the ground as well, so. 327 00:18:57,920 --> 00:19:00,720 Speaker 1: And this would not be out of particularly highlight and 328 00:19:00,759 --> 00:19:01,039 see it. 329 00:19:01,440 --> 00:19:05,079 Speaker 2: You know, we're talking about you know, maybe being around 330 00:19:05,599 --> 00:19:11,000 six to eight hundred kilometers above the Earth, and light 331 00:19:11,079 --> 00:19:16,720 travels three hundred kilometers per millisecond, so that's it's about, you. 332 00:19:16,680 --> 00:19:19,359 Speaker 1: Know, three milliseconds away. It's not not very far. 333 00:19:20,119 --> 00:19:22,559 Speaker 3: Won't worry about that too much though somethings. 334 00:19:23,000 --> 00:19:25,000 Speaker 2: I think it's gonna be some like high late and 335 00:19:25,039 --> 00:19:27,640 sy them like, yeah, no, speed of light. 336 00:19:27,599 --> 00:19:31,759 Speaker 1: Moves pretty fast, moves pretty fast. It's all one. Yeah. 337 00:19:31,920 --> 00:19:34,480 Speaker 4: I think The cool thing also is the the radiators 338 00:19:34,519 --> 00:19:38,240 themselves are about the same size as the existing solar 339 00:19:38,279 --> 00:19:39,720 race for a V three vehicle. 340 00:19:40,960 --> 00:19:42,559 Speaker 1: Kind of kind of in that that realm where we're 341 00:19:42,559 --> 00:19:43,160 flying today. 342 00:19:43,440 --> 00:19:45,680 Speaker 3: Yeah, So I mean they got they got about a 343 00:19:45,759 --> 00:19:49,119 seventy meter wingspan. So these are fairly large, and we're 344 00:19:49,119 --> 00:19:52,119 talking about building a lot of them and putting them 345 00:19:52,200 --> 00:19:56,119 up there. But you like to say like spaces in 346 00:19:56,160 --> 00:19:58,680 the name, Like there's there's a lot of space up there. 347 00:19:58,759 --> 00:20:02,960 And so even when you're talking thousands or even you know, 348 00:20:03,039 --> 00:20:05,799 up to a million satellites, you got plenty of room 349 00:20:05,839 --> 00:20:06,759 to move around up there. 350 00:20:06,880 --> 00:20:10,000 Speaker 2: Yeah, space is really big, So it's like it's not 351 00:20:10,079 --> 00:20:11,440 like space is going to get crowded. 352 00:20:13,160 --> 00:20:14,559 Speaker 1: Space is enormous. 353 00:20:14,559 --> 00:20:16,680 Speaker 2: Like if you're zoom in close to the satellite, looks big, 354 00:20:17,119 --> 00:20:20,160 but if you actually look at it relatively relative to 355 00:20:20,240 --> 00:20:22,359 the Earth, the satellites are so tiny you can you 356 00:20:22,400 --> 00:20:23,599 can't even see them. 357 00:20:24,440 --> 00:20:28,519 Speaker 1: So they're very very tiny compared to Earth. 358 00:20:29,599 --> 00:20:33,039 Speaker 3: And I mean, we have ten about ten thousand starlinks 359 00:20:33,079 --> 00:20:36,240 in orbit right now. We've got a pretty good idea 360 00:20:36,279 --> 00:20:39,960 of how to operate just really large constellations and do 361 00:20:40,039 --> 00:20:40,559 it safely. 362 00:20:40,599 --> 00:20:43,400 Speaker 4: Now right we are the only operator that has any 363 00:20:43,400 --> 00:20:47,519 experience of that scale. It's it's a great thing that 364 00:20:48,160 --> 00:20:49,839 you know we have this background, so we know how 365 00:20:49,880 --> 00:20:52,759 tightly we can pack the satellites and inside them safely. 366 00:20:52,839 --> 00:20:55,319 That's that's a number one goal when we look at 367 00:20:55,359 --> 00:20:55,960 the constellation. 368 00:20:56,559 --> 00:20:59,079 Speaker 3: We're going to be building a lot of satellites, and 369 00:20:59,160 --> 00:21:02,960 we're gonna be building here in Bastra, right, so we've 370 00:21:03,160 --> 00:21:06,920 we've got this, which so we're in the building kind 371 00:21:06,920 --> 00:21:09,079 of in the middle, which we're sitting in that building 372 00:21:09,160 --> 00:21:11,559 right now. This is my first time here. The building 373 00:21:11,640 --> 00:21:14,759 is massive, Like you come around the corner, you see 374 00:21:14,759 --> 00:21:17,119 it through the trees and you're like, oh wow, but 375 00:21:17,599 --> 00:21:19,839 we're about to kind of put this building to shame, 376 00:21:19,839 --> 00:21:20,279 aren't we. 377 00:21:21,440 --> 00:21:25,720 Speaker 2: Yes, we're gonna. In fact, we're already have the solar 378 00:21:25,839 --> 00:21:32,599 manufacturing facility. It's under construction already, and and then we 379 00:21:32,640 --> 00:21:35,799 will be building out the a SAT production building soon. 380 00:21:37,000 --> 00:21:42,039 And yeah, so we expect to have the the ASAT production, 381 00:21:42,240 --> 00:21:48,799 the solo production, and all of that operating at some 382 00:21:48,920 --> 00:21:50,680 reasonable volume by the end of next year. 383 00:21:51,240 --> 00:21:54,599 Speaker 3: So if anybody wants to work on AI satellite, this 384 00:21:54,640 --> 00:21:56,559 is kind of going to become the hub of that. 385 00:21:56,920 --> 00:21:58,759 We're also so I mean, like right behind us, the 386 00:21:58,799 --> 00:22:01,839 machines are humming and we're still making all of our 387 00:22:01,920 --> 00:22:06,079 user terminals for starlink here, that's not going anywhere. In fact, 388 00:22:06,119 --> 00:22:10,039 we're turning on new production lines for new units right Yes. 389 00:22:10,839 --> 00:22:14,680 Speaker 2: In fact, these are the new stalling terminals, which we 390 00:22:14,839 --> 00:22:19,480 made in much high volume than the current terminals, and 391 00:22:19,640 --> 00:22:22,319 ultimately we think there's probably going to be a few one. 392 00:22:22,279 --> 00:22:24,200 Speaker 1: Hundred million stalling terminals out there. 393 00:22:24,640 --> 00:22:29,599 Speaker 2: And then the stalling direct to sell constellation will connect 394 00:22:29,640 --> 00:22:33,759 directed to people's cell phones and enable high vanterwith communication 395 00:22:33,880 --> 00:22:35,519 directly from your phone to space. 396 00:22:35,720 --> 00:22:39,720 Speaker 3: All right, we're two limiting factors down. We've got master orbit, 397 00:22:40,359 --> 00:22:43,839 got putting solar, and a few third ones chips. 398 00:22:44,359 --> 00:22:48,960 Speaker 2: Yes, so at least in the beginning, we can obviously 399 00:22:49,000 --> 00:22:53,720 launch the chips that are already being made. So our 400 00:22:53,759 --> 00:22:59,440 current reference design is for in Vidio Reuben chips or 401 00:22:59,640 --> 00:23:02,880 could be either GB three hundred or or Ruben schiffs, 402 00:23:04,720 --> 00:23:09,519 and we'll also have a reference design for TPUs and 403 00:23:09,640 --> 00:23:12,720 essentially you can put up put any any existing shifts 404 00:23:12,720 --> 00:23:19,559 into into orbit. But the current industry seems to be 405 00:23:21,079 --> 00:23:24,880 it seems like it's gonna I don't get to maybe 406 00:23:24,920 --> 00:23:26,119 around one hundred. 407 00:23:25,799 --> 00:23:29,240 Speaker 1: Gigawatts a year of AI computed but. 408 00:23:31,119 --> 00:23:32,960 Speaker 2: That doesn't answer the question of well, how do you 409 00:23:32,960 --> 00:23:34,200 get to a tarrawatt. 410 00:23:34,559 --> 00:23:40,880 Speaker 1: That's why you need the terrafab always looking a step bigger. Yeah, yeah, in. 411 00:23:40,839 --> 00:23:43,880 Speaker 2: Order to get to the next order of magnitude, you 412 00:23:44,000 --> 00:23:47,720 need a gigantic shift factory to give you a sense 413 00:23:47,720 --> 00:23:51,799 of scale here, we expect that the terrafab is going 414 00:23:51,799 --> 00:23:56,160 to be around one hundred million square feet, which is 415 00:23:57,119 --> 00:24:01,359 ten times the size of the a Tela gig factory Texas. 416 00:24:02,799 --> 00:24:06,000 Speaker 3: And what aside from just you know, I'm going to 417 00:24:06,039 --> 00:24:08,440 need starship point to point to get from one end 418 00:24:08,519 --> 00:24:11,519 to the other. Aside from just the size, what's going 419 00:24:11,599 --> 00:24:15,319 to make this unique different from any other chip building 420 00:24:15,359 --> 00:24:16,519 operation on the planet. 421 00:24:16,880 --> 00:24:18,720 Speaker 2: Well, I think over time there's going to be a 422 00:24:18,759 --> 00:24:22,440 lot of technology evolution with the Terrorfab, but fundamentally it's 423 00:24:22,480 --> 00:24:24,160 about scale. So even if there. 424 00:24:24,039 --> 00:24:24,359 Speaker 1: Were no. 425 00:24:26,599 --> 00:24:31,319 Speaker 2: Fundamental technology breakthroughs, yes, and you simply you could simply 426 00:24:31,359 --> 00:24:38,319 scale the existing chip making technology with a lot of difficulty. 427 00:24:38,519 --> 00:24:40,720 Speaker 1: To a terror watt of chip out per year. 428 00:24:41,880 --> 00:24:43,960 Speaker 2: That's if you look at it just from the logic 429 00:24:44,000 --> 00:24:48,680 die standpoint, that's that's equivalent. That's like having a billion 430 00:24:49,359 --> 00:24:52,839 chips per year with a kilowatt per radical, So so 431 00:24:53,240 --> 00:24:57,960 a billion full radical equivalent chips each doing a kilowatt, 432 00:24:58,400 --> 00:24:59,839 and then you're going to need a lot of memory 433 00:25:00,440 --> 00:25:01,240 to go with that. 434 00:25:01,799 --> 00:25:04,960 Speaker 3: A lot of people today even saying orbital data centers 435 00:25:04,960 --> 00:25:06,119 were like a decade away. 436 00:25:07,119 --> 00:25:07,799 Speaker 1: Yeah, I think we. 437 00:25:07,839 --> 00:25:11,119 Speaker 2: Want to try to give people a sense of the 438 00:25:11,160 --> 00:25:15,559 timeframe we at least the timeframe we're aiming for. I mean, 439 00:25:15,640 --> 00:25:17,720 you know, people should take this with a grain of 440 00:25:17,759 --> 00:25:20,559 sult to undergrade, because this is this is just our 441 00:25:20,559 --> 00:25:22,319 best guess. So this is not a this is not 442 00:25:22,359 --> 00:25:25,599 a promisable we'll do. This is what we what we 443 00:25:25,640 --> 00:25:28,039 are going to try to do and think we probably 444 00:25:28,119 --> 00:25:33,640 can do, which is to get to roughly an annualized 445 00:25:33,720 --> 00:25:35,640 rate of a gigawat per year by the end of 446 00:25:35,720 --> 00:25:40,119 next year in terms of space AI compute, and then 447 00:25:40,759 --> 00:25:44,599 aspirationally scale that by an order of magnitude per year, 448 00:25:45,400 --> 00:25:47,559 so in two and a half years, hitting an annualized 449 00:25:47,640 --> 00:25:49,960 rate of ten giga watts a year to space and 450 00:25:50,039 --> 00:25:53,400 two and a half years, maybe one hundred gigawatts. And 451 00:25:53,440 --> 00:25:57,599 then depending upon or progress there is in chip making 452 00:25:58,359 --> 00:25:59,480 in the rest of the world. 453 00:25:59,240 --> 00:26:02,799 Speaker 1: And with the terriff A going beyond that to scale 454 00:26:02,799 --> 00:26:04,279 to a terror work. 455 00:26:04,119 --> 00:26:07,279 Speaker 2: Per year, which is a thousand gable watts, which is 456 00:26:07,319 --> 00:26:10,559 that's wic the current electricity consumption of the United States. 457 00:26:11,079 --> 00:26:12,759 Speaker 1: I think there will be appetite for that. But we'll 458 00:26:12,759 --> 00:26:15,839 see there's a lot of satellites. 459 00:26:16,400 --> 00:26:18,119 Speaker 2: I don't know what it's going to think about, but uh, 460 00:26:19,440 --> 00:26:20,880 maybe do a lot of simulations or something. 461 00:26:20,960 --> 00:26:22,119 Speaker 1: Yeah. 462 00:26:22,200 --> 00:26:27,000 Speaker 3: So after we've you know, working through all the limiting factors, 463 00:26:27,039 --> 00:26:29,680 we've kind of topped out what we can do on Earth, 464 00:26:30,359 --> 00:26:31,680 what is the next step? 465 00:26:32,079 --> 00:26:32,279 Speaker 1: Two? 466 00:26:33,440 --> 00:26:37,559 Speaker 3: Again try and actually notch maybe some percentage points towards 467 00:26:37,640 --> 00:26:39,480 becoming Kardashev level two? 468 00:26:40,119 --> 00:26:44,000 Speaker 2: Why stop there? Why I think small here? Because a 469 00:26:44,079 --> 00:26:47,839 terror what actually is that's not things small? So there 470 00:26:47,920 --> 00:26:52,240 is in order to get to another three ors nine 471 00:26:52,319 --> 00:26:55,799 two two thousand x from a terror work per year. 472 00:26:56,599 --> 00:26:59,680 The only way that we can really see that you 473 00:26:59,720 --> 00:27:03,000 can achieved that is on the Moon with a mass 474 00:27:03,079 --> 00:27:08,720 driver essentially where you do local production of photobole takes 475 00:27:08,839 --> 00:27:12,839 and so and radiators on the moon. Maybe you bring 476 00:27:12,839 --> 00:27:16,680 the chips from Earth, or you could conceivably make the 477 00:27:16,720 --> 00:27:20,880 chips on the moon, and but you need most of 478 00:27:20,880 --> 00:27:23,160 the mass to be made on the Moon, so you 479 00:27:23,160 --> 00:27:25,119 don't have to transport it to the Moon from Earth. 480 00:27:25,759 --> 00:27:28,960 And and then because the Moon has no atmosphere and 481 00:27:29,079 --> 00:27:32,880 only one sixth Earth's gravity, you can get you can 482 00:27:32,920 --> 00:27:36,759 accelerate the AI satellites into deep space without a rocket, 483 00:27:37,160 --> 00:27:41,079 so you can basically shoot them into space using an 484 00:27:41,119 --> 00:27:44,680 electromagnetic gun like a like a rail gun type. I 485 00:27:44,680 --> 00:27:48,240 mean just it's basically linear electric motor as the way 486 00:27:48,240 --> 00:27:48,839 to think about it. 487 00:27:49,160 --> 00:27:50,559 Speaker 3: So I think where you can show people 488 00:27:58,839 --> 00:28:01,720 Speaker 2: Thanks for listening, See you in the next episode.