1 00:00:00,020 --> 00:00:05,120 Josh: On paper, SpaceX's $1.5 trillion IPO would be the biggest in history, 2 00:00:05,520 --> 00:00:06,500 Josh: raising about $30 billion. 3 00:00:06,840 --> 00:00:11,120 Josh: But that's not the crazy part. The crazy part is that even at that price, 4 00:00:11,340 --> 00:00:13,640 Josh: we still believe it's undervalued. 5 00:00:13,760 --> 00:00:17,200 Josh: In fact, grossly undervalued because this all stems from one realization. 6 00:00:17,460 --> 00:00:22,000 Josh: It's turning Starlink into AI data centers in space. And the winner who deploys 7 00:00:22,000 --> 00:00:24,580 Josh: resources the fastest gets to reshape civilization. 8 00:00:24,900 --> 00:00:27,940 Josh: It isn't just SpaceX stock finally available to the public. 9 00:00:27,940 --> 00:00:31,240 Josh: This is Elon using Wall Street's money to build an 10 00:00:31,240 --> 00:00:35,720 Josh: orbital gpu swarm with these new satellites that are 150 kilowatts and starship 11 00:00:35,720 --> 00:00:39,160 Josh: will launch basically a small city's worth of ai power into orbit every single 12 00:00:39,160 --> 00:00:42,560 Josh: flight we're going to talk about all of this um because eventually this leads 13 00:00:42,560 --> 00:00:47,540 Josh: to lunar factories producing ai satellites and if he pulls this off it's going 14 00:00:47,540 --> 00:00:50,740 Josh: to be pretty amazing so in this episode we're going to break down why this ipo exists 15 00:00:50,740 --> 00:00:52,060 Ejaaz: Only because of ai. 16 00:00:52,060 --> 00:00:55,340 Josh: This was not the case prior to this week how the physics actually favored data 17 00:00:55,340 --> 00:00:59,840 Josh: centers in space and why 1.5 trillion dollars may be a rounding error if SpaceX 18 00:00:59,840 --> 00:01:02,780 Josh: wins because, I mean, in this game, Ejaz, it's kind of simple. 19 00:01:02,860 --> 00:01:06,540 Josh: Whoever can deploy compute the fastest, they win the AI race and they probably 20 00:01:06,540 --> 00:01:08,120 Josh: get to rewrite the future of civilization. 21 00:01:08,740 --> 00:01:12,040 Ejaaz: Yeah. I mean, before we get into it, I want to walk everyone through the kind 22 00:01:12,040 --> 00:01:16,240 Ejaaz: of series of events that kind of led to this groundbreaking bit of news. 23 00:01:16,440 --> 00:01:22,160 Ejaaz: So it all started off around Wednesday this week, where two news sources, 24 00:01:22,300 --> 00:01:30,100 Ejaaz: Bloomberg and the information leaked news that SpaceX was potentially going to IPO sometime in 2026. 25 00:01:30,100 --> 00:01:32,960 Ejaaz: But we didn't quite know the valuation at that point. 26 00:01:33,240 --> 00:01:36,540 Ejaaz: And then it broke that it was a $1.5 trillion valuation. 27 00:01:36,800 --> 00:01:41,520 Ejaaz: And the reason why this is such big news is it would technically be the biggest IPO ever. 28 00:01:42,240 --> 00:01:45,300 Ejaaz: And then people were like, but Elon hasn't confirmed it. 29 00:01:45,480 --> 00:01:50,520 Ejaaz: Well, yesterday, Elon officially confirmed it. He responded to a great article 30 00:01:50,520 --> 00:01:55,720 Ejaaz: from Eric Berger where he kind of predicted why he thinks SpaceX is definitely going to IPO soon. 31 00:01:56,160 --> 00:01:58,920 Ejaaz: And Elon confirming it all but means that, 32 00:01:59,470 --> 00:02:06,210 Ejaaz: If SpaceX IPOs, Josh, at $1.5 trillion, his net worth would effectively double, 33 00:02:06,810 --> 00:02:13,030 Ejaaz: which would put him just under, not exactly at, but just under $1 trillion overnight. 34 00:02:13,290 --> 00:02:16,830 Ejaaz: Once the IPO goes live, he owns 42% of SpaceX. 35 00:02:17,530 --> 00:02:20,510 Ejaaz: So listen, there's a bunch of big numbers here. 36 00:02:20,670 --> 00:02:25,050 Ejaaz: In order to successfully IPO at this valuation, he needs to raise $30 billion. 37 00:02:25,050 --> 00:02:27,690 Ejaaz: Dollars um he needs to you know 38 00:02:27,690 --> 00:02:31,070 Ejaaz: technically deliver on a lot of stuff that we're going to get into in a second 39 00:02:31,070 --> 00:02:36,650 Ejaaz: but you know massive massive vision so the question i ask myself is why now 40 00:02:36,650 --> 00:02:40,590 Ejaaz: and what is he doing to justify this what is he telling the people that he's 41 00:02:40,590 --> 00:02:44,930 Ejaaz: raising 30 billion dollars from um to convince them to give him that kind of 42 00:02:44,930 --> 00:02:47,230 Ejaaz: money and it comes to one thing josh, 43 00:02:47,850 --> 00:02:54,870 Ejaaz: AI data centers in space. Now, you and I have had a very tumultuous relationship with this. 44 00:02:55,230 --> 00:02:56,410 Josh: We've come a long way. 45 00:02:56,550 --> 00:02:59,630 Ejaaz: We've come a very long way. And by a long way, in the last four weeks, 46 00:02:59,790 --> 00:03:03,710 Ejaaz: when this kind of like entire trend started developing, at least on social media. 47 00:03:04,130 --> 00:03:09,870 Ejaaz: And the concept here is if you're able to put GPUs or AI data centers in space 48 00:03:09,870 --> 00:03:13,870 Ejaaz: and harness the energy of the sun, you effectively have infinite power. 49 00:03:13,870 --> 00:03:18,110 Ejaaz: And in this game of creating the best AI to reach AGI level, 50 00:03:18,330 --> 00:03:22,830 Ejaaz: you need as much compute as you can get your hands on. And the fact is Earth's resources is scarce. 51 00:03:23,050 --> 00:03:29,290 Ejaaz: So Elon's whole pitch behind IPO-ing SpaceX is he believes he can control and 52 00:03:29,290 --> 00:03:33,210 Ejaaz: build data centers in space using his Starlink satellite network. 53 00:03:33,350 --> 00:03:37,550 Ejaaz: Josh, he thinks he can create a constellation of Starlink satellites that can 54 00:03:37,550 --> 00:03:39,670 Ejaaz: speak to each other via laser beams, right? 55 00:03:39,810 --> 00:03:43,490 Ejaaz: So they can communicate and share compute and data between themselves, 56 00:03:43,970 --> 00:03:48,470 Ejaaz: harness solar energy from the sun and beam that down to earth to create AGI. 57 00:03:49,350 --> 00:03:52,450 Ejaaz: This sounds like something out of a sci-fi novel, Josh. 58 00:03:53,310 --> 00:03:56,670 Ejaaz: It seems like you're obviously bullish. How feasible is this? 59 00:03:56,750 --> 00:03:57,890 Ejaaz: What are we looking at here? 60 00:03:58,690 --> 00:04:03,530 Josh: Yeah, it's funny because for the longest time, Elon was strictly no. 61 00:04:03,750 --> 00:04:07,030 Josh: No way are we ever going to go public. But the second that they realized that 62 00:04:07,030 --> 00:04:11,310 Josh: Starlink satellites can be architected as this distributed network of data centers, 63 00:04:11,490 --> 00:04:14,090 Josh: it went from no way to we kind of have to. 64 00:04:14,490 --> 00:04:18,470 Josh: Because a lot of this AI race, it kind of comes down to deploying assets just 65 00:04:18,470 --> 00:04:19,590 Josh: quicker than your competitors. 66 00:04:19,810 --> 00:04:24,550 Josh: And we've seen this with XAI in the past where they did not exist, what, two years ago? 67 00:04:24,650 --> 00:04:28,170 Josh: And now they are at the forefront with everybody else. And it's because they 68 00:04:28,170 --> 00:04:30,770 Josh: were able to deploy their resources faster than everyone else. 69 00:04:30,870 --> 00:04:31,930 Josh: They were the fastest to get to 70 00:04:32,490 --> 00:04:35,490 Josh: large coherent cluster of nvidia gpus 71 00:04:35,490 --> 00:04:38,410 Josh: and as a result they have some of the best models in the world so this 72 00:04:38,410 --> 00:04:41,470 Josh: is clearly a race to get to resources spacex has 73 00:04:41,470 --> 00:04:45,630 Josh: the ability to do so xai has the models to do so but spacex doesn't quite have 74 00:04:45,630 --> 00:04:49,770 Josh: the money so their projected revenue was i think it's about 24 billion dollars 75 00:04:49,770 --> 00:04:53,710 Josh: in the near term which for a fun fact is roughly equivalent to nasa's annual 76 00:04:53,710 --> 00:04:57,850 Josh: budget but what they realized is wait a second if we're going to compete on 77 00:04:57,850 --> 00:04:59,730 Josh: a global scale and building AI data centers, 78 00:04:59,870 --> 00:05:01,990 Josh: we need far more money than that. 79 00:05:02,130 --> 00:05:04,190 Josh: And that is the reasoning why they're going to public markets, 80 00:05:04,510 --> 00:05:05,310 Josh: raising the $30 billion. 81 00:05:05,890 --> 00:05:10,450 Josh: And again, at $1.5 trillion, that's about $30 billion they're going to raise in addition. 82 00:05:10,610 --> 00:05:13,750 Josh: So Ijaz, I guess the question, I started by saying this is fairly undervalued 83 00:05:13,750 --> 00:05:15,010 Josh: because the stakes are so high. 84 00:05:15,130 --> 00:05:18,370 Josh: We'll explain the reasoning, but I'm curious what your take is on the valuation, 85 00:05:18,530 --> 00:05:21,330 Josh: the $1.5 trillion number as it sits today. 86 00:05:21,330 --> 00:05:27,290 Ejaaz: Okay, assuming this is economically and physically possible, 87 00:05:27,490 --> 00:05:29,270 Ejaaz: Josh, it's a no-brainer. 88 00:05:29,450 --> 00:05:33,810 Ejaaz: And it's for one simple reason, which Andrew actually outlines pretty well in 89 00:05:33,810 --> 00:05:35,170 Ejaaz: this tweet that I'm showing here. 90 00:05:35,760 --> 00:05:40,860 Ejaaz: To win the AI race, you need the most compute, data, and energy. 91 00:05:41,460 --> 00:05:46,800 Ejaaz: That last one, if you can get infinite energy from the sun, you've already won, right? 92 00:05:47,100 --> 00:05:50,840 Ejaaz: And the only reason why people haven't considered this before is because it hasn't been possible. 93 00:05:51,240 --> 00:05:55,500 Ejaaz: Elon will effectively own the highway and the tolling system into space. 94 00:05:55,640 --> 00:05:58,620 Ejaaz: SpaceX is like literally a decade ahead of the next nearest competitor. 95 00:05:58,840 --> 00:06:02,800 Ejaaz: So if he can create a monopoly on this and actually make this viable, 96 00:06:03,160 --> 00:06:08,160 Ejaaz: he wins. Your point around him needing the money to apply resources to this, 97 00:06:08,280 --> 00:06:11,920 Ejaaz: we all know that Elon started SpaceX to get to Mars, right? 98 00:06:12,000 --> 00:06:15,440 Ejaaz: He wants human civilization on Mars. 99 00:06:15,620 --> 00:06:18,940 Ejaaz: And he didn't really have a good justification as to why he would IPO SpaceX, 100 00:06:19,280 --> 00:06:23,620 Ejaaz: aside from maybe getting extra cash to help fuel that vision. 101 00:06:23,620 --> 00:06:28,640 Ejaaz: Now he realizes that he needs to kind of create this constellation of satellites 102 00:06:28,640 --> 00:06:32,480 Ejaaz: to pay for the Voyagers and the expensive mission to get to Mars. 103 00:06:32,600 --> 00:06:35,540 Ejaaz: So it's a means to an end is the way that I'm kind of thinking about it. 104 00:06:35,900 --> 00:06:42,660 Ejaaz: And so now when I connect the dots, Josh, is $1.5 trillion valuation undervalued or overvalued? 105 00:06:42,760 --> 00:06:47,540 Ejaaz: I think it's undervalued assuming he can get the data centers out in space and 106 00:06:47,540 --> 00:06:48,600 Ejaaz: beam it down back to Earth. 107 00:06:48,660 --> 00:06:52,980 Ejaaz: Because it's not just going to be him and Grok and XAI that benefits from this. 108 00:06:52,980 --> 00:06:56,900 Ejaaz: It's going to be all the other frontier AI labs that are going to want to or 109 00:06:56,900 --> 00:07:00,600 Ejaaz: need to rather use his infrastructure. And that's my simple thesis. 110 00:07:01,300 --> 00:07:04,000 Josh: Yeah, you can ask the question, well, one, like, is this a monopoly? 111 00:07:04,320 --> 00:07:08,880 Josh: Yes, this is a monopoly. The next closest company is five to 10 years behind being Blue Origin. 112 00:07:09,060 --> 00:07:14,720 Josh: Is this a monopoly on a market that is meaningfully large? Yes. 113 00:07:15,200 --> 00:07:18,700 Josh: AI, as well as low earth orbit satellites, as well as energy satellites, 114 00:07:18,860 --> 00:07:21,620 Josh: are about some of the biggest markets on planet earth so the total active 115 00:07:21,620 --> 00:07:24,420 Josh: market size is gigantic and the 116 00:07:24,420 --> 00:07:27,320 Josh: third question is is there an operator capable of actually doing 117 00:07:27,320 --> 00:07:30,360 Josh: the impossible and making the impossible a reality and the 118 00:07:30,360 --> 00:07:33,100 Josh: answer to all three of the questions are yes so there is 119 00:07:33,100 --> 00:07:36,080 Josh: a unique monopoly in a industry that 120 00:07:36,080 --> 00:07:38,900 Josh: is larger than any industry that exists today run by 121 00:07:38,900 --> 00:07:42,760 Josh: the best operator in planet earth and the convergence of those three things if 122 00:07:42,760 --> 00:07:48,480 Josh: it works in the world that we do actually get ai data centers in space and we're 123 00:07:48,480 --> 00:07:50,940 Josh: building artificial general intelligence out there we're harnessing a lot of 124 00:07:50,940 --> 00:07:55,400 Josh: power of the earth that is a civilizational scale shift and the amount of wealth 125 00:07:55,400 --> 00:07:58,580 Josh: generated from that will make 1.5 trillion dollars feel like a drop in the bucket 126 00:07:58,580 --> 00:08:01,880 Josh: so maybe today 1.5 trillion feels large but in the scale of things 127 00:08:02,250 --> 00:08:03,350 Josh: that's a very small number. 128 00:08:03,470 --> 00:08:06,250 Ejaaz: Typically, we speak about companies going from zero to one, and that's where 129 00:08:06,250 --> 00:08:08,670 Ejaaz: those exponential valuations come from. 130 00:08:09,370 --> 00:08:14,570 Ejaaz: With this IPO and with Elon's kind of successful Starship launches and satellite 131 00:08:14,570 --> 00:08:19,090 Ejaaz: launches, we're somewhere between zero and one. We're at 0.5. 132 00:08:19,390 --> 00:08:23,990 Ejaaz: So the risk has been reduced because he's reduced the cost of going to space. 133 00:08:24,150 --> 00:08:29,090 Ejaaz: He's already launched satellites successfully. So now the exponential returns 134 00:08:29,090 --> 00:08:30,790 Ejaaz: seem really, really more feasible. 135 00:08:30,970 --> 00:08:34,430 Ejaaz: So I think he's going to raise that $30 billion in a matter of, 136 00:08:34,430 --> 00:08:36,750 Ejaaz: you know, months, if that amount, right? 137 00:08:37,210 --> 00:08:41,630 Ejaaz: The other thing is, like, he's literally turning science fiction into reality here. 138 00:08:41,850 --> 00:08:46,150 Ejaaz: We talked about the concept of the Kardashev scale, type two civilization, 139 00:08:46,470 --> 00:08:51,010 Ejaaz: which is, you know, humans being able to harness the energy of their local star, that is the sun. 140 00:08:51,190 --> 00:08:53,530 Ejaaz: And we've suddenly gone from it being zero. 141 00:08:53,810 --> 00:08:55,990 Ejaaz: So like that, there's no way he's going to pull it off to, ah, 142 00:08:56,130 --> 00:08:57,530 Ejaaz: you know what? He actually might. 143 00:08:57,810 --> 00:09:02,710 Ejaaz: And so $1.5 trillion seems kind of very undervalued if you have that kind of mindset. 144 00:09:02,970 --> 00:09:05,290 Ejaaz: But again, people listening to the show, Josh, are going to be like, 145 00:09:05,390 --> 00:09:06,870 Ejaaz: well, I don't think the numbers make sense. 146 00:09:07,110 --> 00:09:10,510 Ejaaz: You can't get rid of heat in space. It's a vacuum. 147 00:09:10,710 --> 00:09:13,590 Ejaaz: How are you going to be able to launch satellites? They're going to need to 148 00:09:13,590 --> 00:09:15,430 Ejaaz: be kilometers square wide. 149 00:09:15,870 --> 00:09:18,810 Ejaaz: Josh, have you got any of the numbers that can help break this down? 150 00:09:18,930 --> 00:09:22,150 Josh: Let's run the numbers. So we talked about a good bit of the logistics and 151 00:09:22,150 --> 00:09:24,950 Josh: the cooling in a previous episode earlier this or last week so 152 00:09:24,950 --> 00:09:27,670 Josh: go listen to that if you're interested in that the numbers that 153 00:09:27,670 --> 00:09:30,770 Josh: we're going to talk about today are the amount of scale 154 00:09:30,770 --> 00:09:33,690 Josh: they're actually able to get into outer space and what that looks like so elon 155 00:09:33,690 --> 00:09:37,230 Josh: recently posted on x their ability to get 150 kilowatts 156 00:09:37,230 --> 00:09:42,550 Josh: per satellite into outer space so for some reference numbers the average u.s 157 00:09:42,550 --> 00:09:48,390 Josh: home uses about 10 500 kilowatt hours which is about 1.2 kilowatts of continuous 158 00:09:48,390 --> 00:09:52,410 Josh: energy so if an ai satellite can generate around 150 kilowatt of electrical 159 00:09:52,410 --> 00:09:54,010 Josh: power, like Elon is saying, 160 00:09:54,470 --> 00:10:00,190 Josh: 150 kilowatts divided by 1.2 is about 125 homes worth of continuous power. 161 00:10:00,350 --> 00:10:05,910 Josh: So one of these AI satellites has roughly the power budget of 120 to 130 American 162 00:10:05,910 --> 00:10:07,750 Josh: homes running all the time. 163 00:10:07,930 --> 00:10:12,750 Josh: Now, how much can you do per launch? Because a launch consists of many of these satellites. 164 00:10:13,050 --> 00:10:19,910 Josh: So SpaceX's own site says Starship is designed to carry up to 150 metric tons of low Earth orbit. 165 00:10:20,010 --> 00:10:24,530 Josh: So how many is that? that's about 50 to 75 of these satellites per flight. 166 00:10:24,790 --> 00:10:29,090 Josh: So if you do that math, 60 satellites times 150 kilowatts of energy is nine 167 00:10:29,090 --> 00:10:32,750 Josh: megawatts of power in orbit from one starship, 168 00:10:32,890 --> 00:10:37,970 Josh: which is equivalent to a small town in New England with every single launch 169 00:10:37,970 --> 00:10:39,830 Josh: that they make into outer space. 170 00:10:39,990 --> 00:10:43,430 Josh: That is 7,500 homes. That is 19,000 people's home usage. 171 00:10:43,610 --> 00:10:46,870 Josh: So every fully loaded starship launch is like dropping a small city's worth 172 00:10:46,870 --> 00:10:50,410 Josh: of electric households dedicated just to AI into orbit. 173 00:10:50,650 --> 00:10:55,050 Josh: And that is an outrageously large amount of energy and compute that they're 174 00:10:55,050 --> 00:10:58,550 Josh: putting into outer space, especially relative to what people are trying to do back here on Earth. 175 00:10:58,670 --> 00:11:03,050 Josh: If you remember a Blackwell chip, which is NVIDIA's newest cutting edge rack, 176 00:11:03,230 --> 00:11:05,150 Josh: it uses 120 kilowatts per rack. 177 00:11:05,370 --> 00:11:10,390 Josh: And each one of these satellites surpasses a Blackwell. So it's like this gargantuan 178 00:11:10,390 --> 00:11:13,490 Josh: amount of energy and compute that's being sent to orbit every single launch. 179 00:11:13,750 --> 00:11:18,810 Ejaaz: Yeah, he's tapped into the infinite energy clip and it's become a reality. 180 00:11:18,810 --> 00:11:21,990 Ejaaz: And if he can harness all of that, he basically wins. 181 00:11:22,890 --> 00:11:24,830 Ejaaz: The other thing I was thinking about, Josh, is 182 00:11:25,120 --> 00:11:30,580 Ejaaz: To your point around SpaceX or Starlink specifically being a monopoly on this, 183 00:11:31,060 --> 00:11:33,080 Ejaaz: that's effectively the way that I see it. 184 00:11:33,880 --> 00:11:39,600 Ejaaz: Like pretty much every single usable satellite out there is a Starlink-affiliated satellite. 185 00:11:39,720 --> 00:11:42,820 Ejaaz: In fact, we've got this really cool graphic that we want to show everyone here. 186 00:11:43,100 --> 00:11:44,240 Josh: Oh, yeah, this is sick. 187 00:11:44,540 --> 00:11:49,200 Ejaaz: Okay, if you're looking at this, if you squint, it might look like a bacteriophage, 188 00:11:49,360 --> 00:11:55,080 Ejaaz: right? But no, this is a simulation of all live satellites that are orbiting Earth right now. 189 00:11:55,220 --> 00:11:59,640 Ejaaz: And if you notice my cursor, whenever I touch pretty much any satellite that's 190 00:11:59,640 --> 00:12:04,600 Ejaaz: kind of bumbling around here, you'll see Starlink repeated over and over and over and over again. 191 00:12:04,840 --> 00:12:08,860 Ejaaz: So he has the market monopoly. I think something like 90% of satellites out 192 00:12:08,860 --> 00:12:14,580 Ejaaz: there are Starlink based or have been launched from SpaceX infrastructure and launches. 193 00:12:14,800 --> 00:12:18,540 Ejaaz: So he has the market monopoly. And the reason why that is so important is... 194 00:12:19,230 --> 00:12:24,650 Ejaaz: The capex cost of launching this out into space and making AI data centers in 195 00:12:24,650 --> 00:12:28,390 Ejaaz: space feasible, it actually becomes affordable to do this, right? 196 00:12:28,530 --> 00:12:34,090 Ejaaz: Effectively, if you imagine having just empty data center slots bumbling around 197 00:12:34,090 --> 00:12:38,150 Ejaaz: in space, and all you need to do is fit in the latest GPU and launch it up there 198 00:12:38,150 --> 00:12:40,570 Ejaaz: for cheaper than the cost of the weight of the GPU, 199 00:12:41,630 --> 00:12:43,550 Ejaaz: Elon will be able to do that via SpaceX. 200 00:12:43,790 --> 00:12:48,270 Ejaaz: So it's a crazy monopoly. I do not see anyone coming even close maybe it's uh 201 00:12:48,270 --> 00:12:52,530 Ejaaz: amazon or rather jeff bezos's blue origin but that's still a decade behind right 202 00:12:52,530 --> 00:12:55,230 Ejaaz: so as far as i see it elon's got this in the bag. 203 00:12:55,230 --> 00:12:58,030 Josh: Yeah and they're i mean because of the rapid reusability of 204 00:12:58,030 --> 00:13:02,650 Josh: starship you're able to send just tons of mass to orbit elon he posted that 205 00:13:02,650 --> 00:13:07,970 Josh: he's expecting one megaton per year um so i mean 100 gigawatts it's on the that 206 00:13:07,970 --> 00:13:11,250 Josh: includes 100 gigawatts it's on the order of like a dozen large power plants 207 00:13:11,250 --> 00:13:14,450 Josh: so this map that we're looking at right now full of all these satellites is 208 00:13:14,450 --> 00:13:15,350 Josh: going to get much more full. 209 00:13:15,490 --> 00:13:19,550 Josh: Another interesting thing about this map is the way that they communicate to each other, EJAS. 210 00:13:19,630 --> 00:13:23,550 Josh: A lot of the communication done on Earth is done through fiber optic cables 211 00:13:23,550 --> 00:13:26,970 Josh: when it relates to AI because it's the fastest way to move information as close 212 00:13:26,970 --> 00:13:27,910 Josh: to the speed of light as possible. 213 00:13:28,270 --> 00:13:30,990 Josh: The cool thing about Earth is it's vacuum with no resistance or friction. 214 00:13:31,190 --> 00:13:34,690 Josh: So when these satellites are communicating with each other, even over a distance, 215 00:13:34,950 --> 00:13:37,110 Josh: the only limitation is actually the speed of light. 216 00:13:37,410 --> 00:13:40,650 Josh: And the bandwidth with is huge because there is no friction 217 00:13:40,650 --> 00:13:43,590 Josh: in communication because they are simply sending photons to 218 00:13:43,590 --> 00:13:47,150 Josh: each other they're sending and receiving photons so the unlock that 219 00:13:47,150 --> 00:13:50,770 Josh: that enables where now you have a three-dimensional plane 220 00:13:50,770 --> 00:13:53,570 Josh: of earth where earth is two-dimensional you can only build so much here you 221 00:13:53,570 --> 00:13:56,830 Josh: turn them to three dimensions and you're restrained only by the speed of light 222 00:13:56,830 --> 00:14:01,370 Josh: in terms of photons moving back and forth to each other this gets really sci-fi-esque 223 00:14:01,370 --> 00:14:05,210 Josh: very quickly but also you could start to see how effective this is and what 224 00:14:05,210 --> 00:14:10,090 Josh: an incredible business oh my god i am stoked for the CEO. Are you an investor at $1.5 trillion? 225 00:14:10,910 --> 00:14:13,110 Ejaaz: Yeah. Yeah, of course I am. Yeah. 226 00:14:13,330 --> 00:14:13,790 Josh: Hell yeah. 227 00:14:14,390 --> 00:14:21,010 Ejaaz: Dude, the TAM, the total addressable market, is literally infinite space. 228 00:14:21,530 --> 00:14:22,830 Josh: Yeah, but it does not exist. 229 00:14:23,210 --> 00:14:28,510 Ejaaz: It doesn't exist. Yeah, exactly. Earth will eventually become energy constrained. 230 00:14:28,650 --> 00:14:33,790 Ejaaz: If you look at the energy bills, Josh, of the local towns surrounding data centers 231 00:14:33,790 --> 00:14:38,090 Ejaaz: that are in Abilene, Texas, it's gone through the freaking roof, dude. 232 00:14:38,390 --> 00:14:42,010 Ejaaz: And this is going to be something that we constantly come at odds with because, 233 00:14:42,210 --> 00:14:43,590 Ejaaz: you know, who needs it more? 234 00:14:43,750 --> 00:14:47,530 Ejaaz: What's more important, humans that are kind of local and conducting their own 235 00:14:47,530 --> 00:14:53,570 Ejaaz: kind of business or creating absolute artificial super intelligence for the world to use? 236 00:14:54,090 --> 00:14:56,870 Ejaaz: I know some people might, you know, argue the latter. 237 00:14:57,110 --> 00:15:01,390 Ejaaz: So the point is, we need to find energy elsewhere. And the most obvious one 238 00:15:01,390 --> 00:15:04,910 Ejaaz: is the thing that stares at us in a blue sky every single day. 239 00:15:05,050 --> 00:15:10,150 Ejaaz: So to round this off, Josh, and for the listeners and watchers of this show, 240 00:15:10,310 --> 00:15:12,690 Ejaaz: if you want to imagine what the end game of this looks like, 241 00:15:13,150 --> 00:15:15,850 Ejaaz: it's this graphic that we're showing you on this screen right now, 242 00:15:15,990 --> 00:15:19,130 Ejaaz: except it's Starlink satellites all over the sun. 243 00:15:19,310 --> 00:15:24,470 Ejaaz: And it's just absorbing the energy and transmuting that down to build artificial 244 00:15:24,470 --> 00:15:27,510 Ejaaz: general intelligence on Earth. It's a crazy, crazy vision. 245 00:15:27,970 --> 00:15:30,790 Josh: Things are getting really sci-fi really quick. 246 00:15:31,110 --> 00:15:35,650 Ejaaz: And it's... Josh, I had one more. I had one more thing. Just a little nugget, a little Easter egg. 247 00:15:36,070 --> 00:15:39,490 Ejaaz: Okay. If SpaceX is one of your favorite, AI companies. 248 00:15:40,890 --> 00:15:44,530 Ejaaz: Give me another one. Just give me maybe your next three or four. 249 00:15:44,950 --> 00:15:47,370 Ejaaz: What's one company that comes to mind? I want you to say Google, 250 00:15:47,450 --> 00:15:50,430 Ejaaz: by the way. So you can cut this and say it. 251 00:15:51,610 --> 00:15:55,430 Josh: Well, clearly Google is one of them. We are huge fanboys. We love Google. 252 00:15:55,630 --> 00:15:57,930 Ejaaz: Why do you ask that question? I was hoping you were going to say that. 253 00:15:58,550 --> 00:16:05,170 Ejaaz: They are an official percentage owner in SpaceX to the tune of around 8%. 254 00:16:05,170 --> 00:16:11,110 Ejaaz: That means they invested $900 million in SpaceX about six years ago. 255 00:16:11,750 --> 00:16:20,590 Ejaaz: Josh, that is currently worth $111 billion if SpaceX IPO is at 1.5. 256 00:16:20,590 --> 00:16:23,030 Josh: Trillion. When you got it, you got it, man. And Google just got the juice. 257 00:16:23,330 --> 00:16:24,890 Josh: Listen, I'm a SpaceX shareholder too. 258 00:16:24,990 --> 00:16:30,430 Josh: My friend John and I, we worked so hard back in 2020 to find an SPV of an SPV 259 00:16:30,430 --> 00:16:31,970 Josh: of an SPV to get shares in. 260 00:16:32,130 --> 00:16:35,150 Josh: And it is now up like 50. I'm so excited to finally have some liquidity. 261 00:16:35,370 --> 00:16:37,990 Josh: Not that it matters. I fully intend to add more, but like, man, 262 00:16:38,150 --> 00:16:39,470 Josh: good time to be an optimist. 263 00:16:39,550 --> 00:16:44,290 Josh: Good time to be an investor in the future because it is getting so weird and so wild. So weird. 264 00:16:44,330 --> 00:16:48,950 Josh: But if you are new here, every Friday, we release an episode about kind of a 265 00:16:48,950 --> 00:16:50,850 Josh: roundup of the world in AI. There's more stuff that happened. 266 00:16:50,970 --> 00:16:53,750 Josh: And this second half of the episode, we're getting into that, 267 00:16:53,870 --> 00:16:56,570 Josh: starting with what feels like a natural extension of this e-jazz, right? Where, 268 00:16:56,930 --> 00:17:00,550 Josh: We officially have the first AI data center in space that actually worked. 269 00:17:01,130 --> 00:17:06,930 Ejaaz: Yep. So this tweet that you're seeing from Philip Johnson, Philip is kind of 270 00:17:06,930 --> 00:17:09,430 Ejaaz: like the guy that kicked off this trend, I would argue. 271 00:17:09,930 --> 00:17:17,370 Ejaaz: He is the CEO and founder of StarCloud, which is a startup incubated by Y Combinator 272 00:17:17,370 --> 00:17:20,730 Ejaaz: with a very bold vision of putting AI data centers in space. 273 00:17:21,050 --> 00:17:26,530 Ejaaz: The one difference between Philip and Elon is he's actually put a data center 274 00:17:26,530 --> 00:17:30,930 Ejaaz: in space and he launched it through one of Elon's Falcon rockets last month. 275 00:17:31,410 --> 00:17:37,470 Ejaaz: And they put an NVIDIA H100 in space. It's currently out there running inference and training models. 276 00:17:37,650 --> 00:17:40,830 Ejaaz: And it was able to do two really cool things, Josh. 277 00:17:41,210 --> 00:17:46,950 Ejaaz: Number one, it trained a version, an open source version of ChatGPT from scratch 278 00:17:46,950 --> 00:17:48,790 Ejaaz: on all of Shakespeare's works. 279 00:17:49,330 --> 00:17:52,930 Ejaaz: Just for fun, to see if it could figure it out. And secondly, 280 00:17:53,190 --> 00:17:58,770 Ejaaz: it took an open source version of Google's AI model called Gemma and is currently, 281 00:17:59,030 --> 00:18:03,370 Ejaaz: right as we speak, inferencing between that model in outer space back to Earth. 282 00:18:03,470 --> 00:18:07,290 Ejaaz: So you can send a prompt to the Gemma model out there and get a response in 283 00:18:07,290 --> 00:18:09,890 Ejaaz: a couple of seconds, which is just insane to say. 284 00:18:10,030 --> 00:18:13,050 Ejaaz: Now, people listening to this might be like, well, an NVIDIA H100, 285 00:18:13,230 --> 00:18:14,450 Ejaaz: that's an old GPU, who cares? 286 00:18:14,570 --> 00:18:17,510 Ejaaz: And it's just one of them. This isn't the vision that you just pitched me of 287 00:18:17,510 --> 00:18:19,570 Ejaaz: a satellite network eclipsing the sun. 288 00:18:20,230 --> 00:18:24,970 Ejaaz: Baby steps guys this is baby steps the fact that this uh gpu can survive out 289 00:18:24,970 --> 00:18:29,710 Ejaaz: there dispels the rumors that you know oh you can't radiate the heat or radiation 290 00:18:29,710 --> 00:18:33,690 Ejaaz: is going to kill the gpu it's it's out there it's been running for two and a 291 00:18:33,690 --> 00:18:37,650 Ejaaz: half weeks now so this is a very feasible thing uh josh are you excited about this. 292 00:18:37,650 --> 00:18:40,470 Josh: I'm i'm excited about it it's a great proof of concept 293 00:18:40,470 --> 00:18:43,910 Josh: it shows like okay shielding works okay heat dissipation 294 00:18:43,910 --> 00:18:46,710 Josh: works what i really love the most about this is the models that 295 00:18:46,710 --> 00:18:50,250 Josh: they run uh here on the post that you're showing they trained the nano 296 00:18:50,250 --> 00:18:53,070 Josh: gpt model from andre carpathy uh to work 297 00:18:53,070 --> 00:18:55,830 Josh: on those shakespeare things so it's cool like our i mean 298 00:18:55,830 --> 00:19:00,130 Josh: number one fanboy of andre here he created this like small little nano gpt and 299 00:19:00,130 --> 00:19:03,210 Josh: now they've created a model in space and there's two models running there there's 300 00:19:03,210 --> 00:19:05,290 Josh: one that they actually trained and then there's one that they had pre-built 301 00:19:05,290 --> 00:19:09,310 Josh: that is returning tokens this is awesome it is a proof of concept and it shows 302 00:19:09,310 --> 00:19:13,450 Josh: that i mean granted i think people have now realized that this is possible but 303 00:19:13,450 --> 00:19:15,310 Josh: they are actively proving it in reality. 304 00:19:15,490 --> 00:19:18,070 Josh: And for that, it's awesome. StarCloud seems to be doing cool stuff. 305 00:19:18,330 --> 00:19:20,170 Josh: I'm very happy with the company, very excited for the future. 306 00:19:21,170 --> 00:19:25,350 Ejaaz: Yeah, but enough talk about GPUs in space, Josh. 307 00:19:25,790 --> 00:19:29,350 Josh: Let's kind of like touch. You mean we're going to ground this conversation? 308 00:19:29,610 --> 00:19:34,430 Ejaaz: I am grounding the conversation. I am touching wood. I don't have any grass 309 00:19:34,430 --> 00:19:36,390 Ejaaz: near me right now, but I'm back on Earth officially. 310 00:19:36,890 --> 00:19:42,470 Ejaaz: There is a GPU war going on between our two favorite foes, Josh. 311 00:19:42,710 --> 00:19:46,950 Ejaaz: The USA, and China. Now, they've had a very tumultuous relationship, 312 00:19:47,290 --> 00:19:52,390 Ejaaz: mainly because NVIDIA has been restricted from selling their GPUs to China. 313 00:19:52,510 --> 00:19:53,250 Josh: Which, by the way, 314 00:19:53,530 --> 00:19:57,530 Ejaaz: Makes up a huge percentage of their revenue every year. So the fact that Trump 315 00:19:57,530 --> 00:20:00,610 Ejaaz: said no to Jensen was a big deal. 316 00:20:01,090 --> 00:20:07,130 Ejaaz: Until this week, when he gave the thumbs up to sell not just any NVIDIA GPU, 317 00:20:07,390 --> 00:20:12,910 Ejaaz: but their second older generation GPU, the H100 or H200 sorry. 318 00:20:13,540 --> 00:20:20,300 Ejaaz: To China. And it comes with a few stipulations, Josh, which is 25% of revenue 319 00:20:20,300 --> 00:20:24,660 Ejaaz: that NVIDIA makes from these sales to China needs to go to the USGov. 320 00:20:24,800 --> 00:20:27,340 Ejaaz: So he's paying a pretty hefty tax on this. 321 00:20:27,640 --> 00:20:31,760 Ejaaz: And so people were pretty excited. NVIDIA stock jumped up on the news, 322 00:20:31,900 --> 00:20:35,640 Ejaaz: except on the same trading session in that same day, 323 00:20:35,880 --> 00:20:39,560 Ejaaz: it retraced because China put out a commentary officially saying, 324 00:20:39,560 --> 00:20:42,460 Ejaaz: we don't want your gpus and in fact we're 325 00:20:42,460 --> 00:20:45,800 Ejaaz: going to restrict the amount of h200 gpus that our 326 00:20:45,800 --> 00:20:48,820 Ejaaz: companies buy from you because they can just use chinese 327 00:20:48,820 --> 00:20:52,080 Ejaaz: gpus right so that was the response from china 328 00:20:52,080 --> 00:20:55,440 Ejaaz: saying like listen we don't need it so it was a very volatile uh 329 00:20:55,440 --> 00:20:59,220 Ejaaz: day in terms of trading the nvidia stock except that 330 00:20:59,220 --> 00:21:02,040 Ejaaz: then the chinese companies themselves spoke up and 331 00:21:02,040 --> 00:21:04,940 Ejaaz: said uh hey we're deep seek over here 332 00:21:04,940 --> 00:21:07,980 Ejaaz: and we kind of need nvidia's gpus 333 00:21:07,980 --> 00:21:10,980 Ejaaz: because we can't trade oh we 334 00:21:10,980 --> 00:21:14,080 Ejaaz: can't train frontier level intelligence unless we 335 00:21:14,080 --> 00:21:19,200 Ejaaz: have those gpus so there's this there's been this big back and forth china in 336 00:21:19,200 --> 00:21:23,760 Ejaaz: response to that has been like okay let's hold like a council session internally 337 00:21:23,760 --> 00:21:28,140 Ejaaz: let me talk to some of the companies here and figure out how many h200s you 338 00:21:28,140 --> 00:21:31,900 Ejaaz: guys actually need and i have a few thoughts about this josh but what's your 339 00:21:31,900 --> 00:21:33,120 Ejaaz: general reaction before. 340 00:21:33,120 --> 00:21:36,480 Josh: We get into it. It's funny. So the story isn't the fact that we made them available 341 00:21:36,480 --> 00:21:39,820 Josh: to China. It's that China said, we don't want them. 342 00:21:40,100 --> 00:21:43,740 Josh: In fact, we already have them. We have stolen them. We have smuggled them in. 343 00:21:43,880 --> 00:21:45,760 Josh: So that is a generous offer of you. 344 00:21:45,920 --> 00:21:49,400 Josh: But hey, we've already been training models on your hardware for a long time. 345 00:21:49,580 --> 00:21:50,820 Josh: So thanks, but no thanks. 346 00:21:51,080 --> 00:21:55,220 Josh: And that comes to the surprise of probably no one. Everyone had an idea that 347 00:21:55,220 --> 00:21:56,180 Josh: a lot of these were getting sent to 348 00:21:59,740 --> 00:22:05,360 Josh: I suspect what that means is these new DeepSeq models, which have been hyper-efficient 349 00:22:05,360 --> 00:22:08,840 Josh: in the past, will just become even more efficient going forward. 350 00:22:08,860 --> 00:22:12,400 Josh: And we've always talked about this constraint that they've had, 351 00:22:12,520 --> 00:22:15,700 Josh: where they've been resource-limited, therefore they've had to be resourceful on the software stack. 352 00:22:15,880 --> 00:22:18,400 Josh: In a world where they are resourceful on the software stack, 353 00:22:18,500 --> 00:22:22,620 Josh: but also have a competitive hardware stack, that feels like a supercharged DeepSeq. 354 00:22:22,700 --> 00:22:28,580 Josh: And I feel like whatever model they release next is going to be a seriously heavy hitter. 355 00:22:29,740 --> 00:22:34,320 Ejaaz: Yeah, I mean, there's, it's a few things. There's the performance competition 356 00:22:34,320 --> 00:22:39,040 Ejaaz: that you just explained, but there's also a strategic political one. 357 00:22:39,180 --> 00:22:42,040 Ejaaz: And the strategic political one has two sides, okay? 358 00:22:42,220 --> 00:22:48,640 Ejaaz: So if I'm China, I don't want my Chinese companies hooked on American GPUs. 359 00:22:48,740 --> 00:22:51,260 Ejaaz: I want them to use Chinese-made GPUs. Why? 360 00:22:51,480 --> 00:22:55,320 Ejaaz: Because if the Chinese-made GPUs are good enough to create frontier level intelligence, 361 00:22:55,320 --> 00:22:59,000 Ejaaz: we now have less of a dependency on the US. 362 00:22:59,180 --> 00:23:02,340 Ejaaz: In fact, it's probably the most important dependency that China has on the US 363 00:23:02,340 --> 00:23:07,800 Ejaaz: right now, stretching across from raw materials to computer energy to whatever you might be, right? 364 00:23:08,020 --> 00:23:12,100 Ejaaz: So they want to rely on Chinese made GPUs, not the US made GPUs. 365 00:23:12,240 --> 00:23:16,140 Ejaaz: On the US side of things, they're thinking of this specific strategy, which is, 366 00:23:16,840 --> 00:23:23,600 Ejaaz: If we can keep selling China, older generation NVIDIA GPUs, their intelligence 367 00:23:23,600 --> 00:23:27,260 Ejaaz: will always lag ours and we will always be the leader. 368 00:23:27,540 --> 00:23:33,260 Ejaaz: The second way to look at that is also NVIDIA has a really sticky software moat. 369 00:23:33,440 --> 00:23:37,680 Ejaaz: So typically anyone that runs NVIDIA GPUs isn't just hooked to them because 370 00:23:37,680 --> 00:23:38,600 Ejaaz: they're using the hardware. 371 00:23:38,760 --> 00:23:42,840 Ejaaz: It's because they rely on the software stack called CUDA, C-U-D-A. 372 00:23:42,840 --> 00:23:45,720 Ejaaz: I think it stands for Compute Unified Device Architecture, right? 373 00:23:46,400 --> 00:23:50,000 Ejaaz: The more and more they use NVIDIA GPUs, the more and more they rely on it, 374 00:23:50,080 --> 00:23:53,320 Ejaaz: Josh. So it's kind of like a drug for them. So this is strategic from the US. 375 00:23:53,480 --> 00:23:56,480 Ejaaz: They're like, sell them the GPUs, just take a 25% stack. 376 00:23:57,220 --> 00:24:02,060 Ejaaz: We're releasing Rubin, which is NVIDIA's latest GPUs coming out in 2026. 377 00:24:02,520 --> 00:24:05,600 Ejaaz: Just keep them hooked on it. Keep them behind. Let's keep them in check. 378 00:24:05,920 --> 00:24:08,880 Josh: Well, it seems like that's what they're doing. So that's the update on the China 379 00:24:08,880 --> 00:24:11,320 Josh: front, which is interesting. But there's another update on the hardware front, 380 00:24:11,420 --> 00:24:12,900 Josh: EGIS, which I'm very excited to talk about, 381 00:24:13,320 --> 00:24:16,380 Ejaaz: And that is Google's new glasses. 382 00:24:16,680 --> 00:24:20,540 Josh: Now, I am old enough to remember the Google Glass version 1, 383 00:24:21,080 --> 00:24:24,100 Josh: which came out probably over a decade now. 384 00:24:24,400 --> 00:24:30,760 Josh: And at the time, it was viewed as the nerdy Silicon Valley kind of like freak 385 00:24:30,760 --> 00:24:34,320 Josh: headwear device, and it never went anywhere. It didn't work very well. 386 00:24:34,480 --> 00:24:38,500 Josh: But all these years later, it appears as if Google has announced that they're 387 00:24:38,500 --> 00:24:41,140 Josh: getting back in the hardware game, back in the eyewear game, 388 00:24:41,140 --> 00:24:42,620 Josh: and releasing these classes. 389 00:24:42,840 --> 00:24:46,320 Josh: So Ejaz, please fill me in. Tell me why these are not going to suffer the same 390 00:24:46,320 --> 00:24:53,860 Josh: fate as Google S 1.0 or Meta's Ray-Bans, which are still not on your face for good reason. 391 00:24:53,860 --> 00:24:58,160 Josh: So it's a very difficult thing to build the good eyewear and no one's done it. 392 00:24:58,380 --> 00:25:01,580 Josh: Why is Google going to make it differently if they even can? 393 00:25:02,410 --> 00:25:07,730 Ejaaz: So in terms of like specific details, we don't have much information beyond 394 00:25:07,730 --> 00:25:10,010 Ejaaz: this demo video that we're showing on screen here. 395 00:25:10,470 --> 00:25:15,590 Ejaaz: They've termed the project Project Aura, which is basically going to be their 396 00:25:15,590 --> 00:25:18,050 Ejaaz: next gen version of Google Class. 397 00:25:18,510 --> 00:25:24,430 Ejaaz: And from my initial reaction from this video is it looks very similar to the 398 00:25:24,430 --> 00:25:29,270 Ejaaz: Apple Vision Pro experience, coupled with a few different features that Meta 399 00:25:29,270 --> 00:25:32,830 Ejaaz: Ray-Bans kind of introduced in their demo a few months earlier. 400 00:25:33,790 --> 00:25:38,970 Ejaaz: My take is it's, okay, number one, it's gonna be better than Google Glass. I would hope so, right? 401 00:25:39,550 --> 00:25:43,090 Ejaaz: From the design of the glass, it looks more casual and applicable to a wider 402 00:25:43,090 --> 00:25:45,870 Ejaaz: range of waters. It looks a little chunky on her head now that I'm actually 403 00:25:45,870 --> 00:25:47,530 Ejaaz: looking at the person that's using this. 404 00:25:47,610 --> 00:25:52,350 Josh: I was gonna say, they look beefy. Those are some chunky glasses. 405 00:25:52,890 --> 00:25:57,430 Ejaaz: That's definitely way too big, which kind of implies that, you know, 406 00:25:57,530 --> 00:25:58,870 Ejaaz: assuming this is a prototype, 407 00:26:00,150 --> 00:26:03,870 Ejaaz: they've got a chunky kind of build-out a physical build-out like the computer 408 00:26:03,870 --> 00:26:07,830 Ejaaz: running I'm super curious what the sensors look like here how they've been able to optimize for it, 409 00:26:08,470 --> 00:26:11,350 Ejaaz: aesthetically it doesn't look like the best but functionally it 410 00:26:11,350 --> 00:26:15,270 Ejaaz: looks pretty cool like again like that Apple Vision Pro experience you can kind 411 00:26:15,270 --> 00:26:18,750 Ejaaz: of like have several desktop screens across you can access all your favorite 412 00:26:18,750 --> 00:26:23,810 Ejaaz: apps the number one thing that I'm excited about this Josh is Google has built 413 00:26:23,810 --> 00:26:28,950 Ejaaz: out a range of different apps and services that I use on a daily basis this. 414 00:26:29,130 --> 00:26:34,310 Ejaaz: So if I can somehow kind of combine that in with my visual experience on a daily 415 00:26:34,310 --> 00:26:36,790 Ejaaz: basis, it automatically becomes useful, right? 416 00:26:37,370 --> 00:26:43,010 Ejaaz: Apple kind of took a step in this direction going from the Apple iPhone to the Apple Watch, right? 417 00:26:43,110 --> 00:26:45,910 Ejaaz: People didn't have to pull out their phone, they could just kind of like look at their wrist. 418 00:26:46,630 --> 00:26:50,350 Ejaaz: Google can now have the benefit of like kind of your vision doing this. So, 419 00:26:50,760 --> 00:26:53,060 Ejaaz: automatically through that, I think it's going to be useful. 420 00:26:53,340 --> 00:26:58,820 Ejaaz: In terms of viability, I'm going to base it heavily on the Meta Ray-Bans reception, 421 00:26:59,040 --> 00:27:00,820 Ejaaz: which was absolutely terrible. 422 00:27:01,160 --> 00:27:05,700 Ejaaz: I think Google has the benefit of being able to scale hardware manufacturing 423 00:27:05,700 --> 00:27:09,920 Ejaaz: way better or have more experience doing that way more than Meta can. 424 00:27:10,100 --> 00:27:13,720 Ejaaz: So that's pretty bullish, but it's going to come down to the execution and I'm 425 00:27:13,720 --> 00:27:15,220 Ejaaz: going to reserve my right to judge for now. 426 00:27:15,820 --> 00:27:18,080 Josh: Yeah, I'll go on record as being the biggest hater for this product. 427 00:27:18,220 --> 00:27:18,920 Josh: I think it's going to be terrible. 428 00:27:19,440 --> 00:27:23,160 Josh: But the thing that I love is that Google's making it. They're working on it. 429 00:27:23,320 --> 00:27:26,020 Josh: And I'm excited for version three of whatever this is. 430 00:27:26,340 --> 00:27:29,500 Josh: So whatever they make this year or next year in 2026, this demo, 431 00:27:29,880 --> 00:27:30,940 Josh: that's not really appealing. 432 00:27:31,180 --> 00:27:32,160 Josh: Whatever the version two is, 433 00:27:32,260 --> 00:27:35,400 Josh: maybe the battery shrinks, maybe the glasses shrink. Still not that great. 434 00:27:35,800 --> 00:27:38,220 Josh: Version three, normally version three of these things start to get good. 435 00:27:38,340 --> 00:27:43,740 Josh: So maybe 2028 will be somewhere in the reasonably viable glasses world. 436 00:27:44,160 --> 00:27:47,280 Josh: But it's cool. I'm glad Google's in the game because this is very clearly an 437 00:27:47,280 --> 00:27:50,600 Josh: important form factor in the future of the way we interface with AI and computers. 438 00:27:50,840 --> 00:27:52,980 Josh: And there's another company now working on manufacturing them. 439 00:27:53,080 --> 00:27:55,320 Josh: So for that, it is a win for everybody. 440 00:27:55,500 --> 00:27:59,820 Ejaaz: And the final item on the docket comes from the open source world. 441 00:28:00,220 --> 00:28:04,960 Ejaaz: Noose Research, which is a frontier open source intelligence lab, 442 00:28:05,180 --> 00:28:06,700 Ejaaz: had a really big breakthrough. 443 00:28:07,000 --> 00:28:09,980 Ejaaz: And by big, I mean really small, big breakthrough. 444 00:28:10,420 --> 00:28:16,060 Ejaaz: They launched a 30 billion parameter model, Josh, which compared to frontier 445 00:28:16,060 --> 00:28:20,980 Ejaaz: intelligence models right now, which range from, I think, 700 billion to 1 trillion 446 00:28:20,980 --> 00:28:23,220 Ejaaz: parameters, this is tiny. 447 00:28:23,540 --> 00:28:31,280 Ejaaz: Except it did a very big thing. It scored an 87 out of 120 on the notorious 448 00:28:31,280 --> 00:28:33,380 Ejaaz: Putnam mathematics competition. 449 00:28:33,920 --> 00:28:37,420 Ejaaz: For those of you who aren't aware of this competition, it is incredibly hard 450 00:28:37,420 --> 00:28:40,600 Ejaaz: for any of the smartest humans in the world to do. 451 00:28:40,600 --> 00:28:45,740 Ejaaz: And in fact, comparing its results to the human scores of last year, 452 00:28:45,900 --> 00:28:50,700 Ejaaz: it would have placed second, scoring a problem set of eight perfect scores, which is just 453 00:28:51,080 --> 00:28:56,300 Ejaaz: crazy. So the fact that this tinier model packs such a crazy punch is nuts. 454 00:28:56,520 --> 00:29:00,720 Ejaaz: Now, for those of you thinking, oh, well, it's just a mathematics nerdy model, who the hell cares? 455 00:29:01,120 --> 00:29:08,040 Ejaaz: Bear in mind that Alibaba spent, I think, upwards of $15 billion in collectively 456 00:29:08,040 --> 00:29:13,560 Ejaaz: training their latest QEN3 model, which is a pretty amazing open source model. 457 00:29:13,740 --> 00:29:18,980 Ejaaz: They scored 27 out of 120 doing the same test. 458 00:29:19,420 --> 00:29:22,800 Ejaaz: So the reason why this excites me is for a few reasons. Number one. 459 00:29:23,360 --> 00:29:25,360 Ejaaz: Noose Research open sourced this entire thing. 460 00:29:25,500 --> 00:29:28,740 Ejaaz: So if you're listening to this and you're curious to kind of test this out yourself, you can do it. 461 00:29:28,840 --> 00:29:32,340 Ejaaz: And 30 billion parameter models is something that you can feasibly run on consumer 462 00:29:32,340 --> 00:29:34,160 Ejaaz: hardware at home. That's number one. 463 00:29:34,560 --> 00:29:37,520 Ejaaz: Number two, they did this in a distributed fashion. 464 00:29:37,800 --> 00:29:41,200 Ejaaz: So typically when you're training a model, you want to build heavy data centers 465 00:29:41,200 --> 00:29:45,200 Ejaaz: and run a bunch of reinforcement learning, reasoning, stuff like that to create 466 00:29:45,200 --> 00:29:46,780 Ejaaz: the model. They didn't use any of that. 467 00:29:46,880 --> 00:29:51,380 Ejaaz: They used a distributed network of compute to be able to create this. So that's really cool. 468 00:29:51,600 --> 00:29:56,440 Ejaaz: And number three, something that they pioneered, Josh, is something called an agent harness. 469 00:29:56,700 --> 00:30:00,840 Ejaaz: Basically, the way that this model became so smart is in the post-training phase, 470 00:30:01,100 --> 00:30:05,820 Ejaaz: where they spun up a number of AI agents, which reasoned with each other in 471 00:30:05,820 --> 00:30:08,920 Ejaaz: a competitive tournament bracket style thing. 472 00:30:09,080 --> 00:30:12,560 Ejaaz: And the one with the best solution won and submitted their answer. 473 00:30:12,680 --> 00:30:14,720 Ejaaz: And that's how they reached this frontier level intelligence. 474 00:30:14,720 --> 00:30:20,360 Ejaaz: The reason why this is so important is if you assume that compute is the only 475 00:30:20,360 --> 00:30:25,820 Ejaaz: thing you need to train frontier-level intelligence, this experiment disproves that. 476 00:30:26,020 --> 00:30:27,660 Ejaaz: This experiment proves to you that 477 00:30:27,660 --> 00:30:30,980 Ejaaz: you can achieve that same level of intelligence with much less compute. 478 00:30:31,180 --> 00:30:34,400 Josh: It's awesome they're getting these distilled models to be so powerful. 479 00:30:34,620 --> 00:30:36,560 Josh: And this is kind of a trend that... 480 00:30:37,130 --> 00:30:40,390 Josh: Projects towards a loose bear case for ai which is if if 481 00:30:40,390 --> 00:30:44,250 Josh: you can continue to distill these down into these hyper-optimized 482 00:30:44,250 --> 00:30:47,250 Josh: models that can eventually run on your phone and does 483 00:30:47,250 --> 00:30:52,490 Josh: the ai at the edge overpower the ai in the data center and does that mean that 484 00:30:52,490 --> 00:30:57,330 Josh: the value of these large data centers goes down i don't think so but this is 485 00:30:57,330 --> 00:31:02,130 Josh: a proof of concept and directionally proving or i guess giving us some data 486 00:31:02,130 --> 00:31:06,030 Josh: on what that actually means and how that looks this is cool It's a fun project. 487 00:31:06,250 --> 00:31:09,170 Josh: And the fact that they're doing it in a decentralized way seems interesting. 488 00:31:09,470 --> 00:31:14,030 Ejaaz: That wraps up the end of today's episode. It was a big one. 489 00:31:14,850 --> 00:31:20,150 Ejaaz: If you can't tell, Josh and I are slight bulls on SpaceX and Elon's companies. 490 00:31:20,330 --> 00:31:22,590 Josh: You didn't show off your hoodie. Show off your hoodie. Oh, I didn't. 491 00:31:22,690 --> 00:31:24,070 Ejaaz: I got my news research hoodie. 492 00:31:24,210 --> 00:31:27,590 Josh: Yeah, so clearly you see where both of us stand here. Yes. Space maximalist. 493 00:31:27,930 --> 00:31:28,190 Ejaaz: Yes. 494 00:31:28,670 --> 00:31:29,890 Josh: Other maximalist. 495 00:31:30,090 --> 00:31:31,090 Ejaaz: Open source. Distributed. 496 00:31:31,670 --> 00:31:33,030 Josh: Distributed. Distributed training. Sick. 497 00:31:34,090 --> 00:31:39,670 Ejaaz: I'm a fan of both. I'm a fan of both. And one thing has become clear to me as 498 00:31:39,670 --> 00:31:45,190 Ejaaz: we wrap up this episode is that the future of AI is dependent on a number of 499 00:31:45,190 --> 00:31:46,630 Ejaaz: different protocol layers, Josh. 500 00:31:47,090 --> 00:31:49,110 Ejaaz: I don't think SpaceX wins without 501 00:31:49,110 --> 00:31:53,510 Ejaaz: the help of Tesla, without the help of Starlink, without the help of XAI. 502 00:31:53,710 --> 00:31:58,170 Ejaaz: They all kind of feed into each other and I'm excited to see how this industry builds out. 503 00:31:59,330 --> 00:32:03,490 Ejaaz: You can bet your ass Limitless is going to be the channel that breaks all this 504 00:32:03,490 --> 00:32:06,590 Ejaaz: news. I just want us to take a little bit of a victory lap, Josh. 505 00:32:07,110 --> 00:32:12,510 Ejaaz: We have called out two very important trends months earlier that have become super important. 506 00:32:12,750 --> 00:32:17,710 Ejaaz: One of them being AI data centers in space, the other one being the Google bull case. Am I missing any? 507 00:32:18,520 --> 00:32:21,900 Josh: Um, well, another one, we had Blake Scholl on the podcast who does supersonic 508 00:32:21,900 --> 00:32:26,160 Josh: jets and he just announced a supersonic jet generator for AI data center. 509 00:32:26,300 --> 00:32:29,880 Josh: So we've been early and we've been right. I mean, granted, directionally right. 510 00:32:29,980 --> 00:32:33,180 Josh: We first started off as haters, but we've haters of the AI space, 511 00:32:33,200 --> 00:32:36,000 Josh: but we've changed our mind and we've been covering it along the way. 512 00:32:36,140 --> 00:32:39,280 Josh: So listen, it might not always be right, but it's always early and you'll always 513 00:32:39,280 --> 00:32:44,260 Josh: be up to date on everything that matters in this world of AI and frontier technology. 514 00:32:44,860 --> 00:32:48,200 Ejaaz: Yes. So if you're listening to this and you aren't subscribed, 515 00:32:48,420 --> 00:32:51,020 Ejaaz: and that is 80% of you, by the way, please subscribe. 516 00:32:51,340 --> 00:32:54,260 Ejaaz: It helps us out so much. Wherever you're listening to it, it could be on Spotify, 517 00:32:54,540 --> 00:32:55,660 Ejaaz: Apple, YouTube, whatever. 518 00:32:55,780 --> 00:32:58,960 Ejaaz: If you're on YouTube, actually hit the notifications button because that also 519 00:32:58,960 --> 00:33:02,800 Ejaaz: alerts you of the latest alpha. We drop three to four episodes a week. It's awesome. 520 00:33:03,080 --> 00:33:09,720 Ejaaz: And the final call to action is we have a banger of a newsletter which drops every Friday. 521 00:33:09,840 --> 00:33:13,360 Ejaaz: And from next week, it drops twice, every Wednesday and Friday, 522 00:33:13,360 --> 00:33:18,020 Ejaaz: one with an SAE investment thesis on the company that Josh or I are super bullish 523 00:33:18,020 --> 00:33:24,200 Ejaaz: about and the other highlighting the top five bits of news in AI and Frontier Tech of that week. 524 00:33:24,320 --> 00:33:25,940 Ejaaz: So you don't want to miss it. If you want to keep up to date, 525 00:33:26,160 --> 00:33:27,660 Ejaaz: subscribe to the Limitless Ecosystem. 526 00:33:27,820 --> 00:33:29,840 Ejaaz: We will see you on the next one.