1 00:00:00,020 --> 00:00:04,340 Ejaaz: This is the most ridiculous turn of events. No one saw this coming. 2 00:00:04,520 --> 00:00:10,420 Ejaaz: Two of the biggest rivals in AI, SpaceX and Anthropic, are teaming up to take 3 00:00:10,420 --> 00:00:13,200 Ejaaz: down Sam Altman at OpenAI. 4 00:00:13,460 --> 00:00:17,040 Ejaaz: And the reason why this is so ridiculous is literally a month ago, 5 00:00:17,240 --> 00:00:19,060 Ejaaz: Elon was posting tweets like this. 6 00:00:19,300 --> 00:00:22,400 Ejaaz: Anthropic hates Western civilization. What else have we got? 7 00:00:22,520 --> 00:00:26,500 Ejaaz: We've got your AI hates white Asians, heterosexuals, and men. 8 00:00:26,660 --> 00:00:31,440 Ejaaz: This is misanthropic and evil. all the way to today or yesterday where he announced 9 00:00:31,440 --> 00:00:36,320 Ejaaz: that I've reviewed Dario, I've spoken to the senior team at Anthropic, 10 00:00:36,580 --> 00:00:39,460 Ejaaz: and I have determined that they are in fact not evil. 11 00:00:39,660 --> 00:00:43,280 Ejaaz: And Claude, I quote, will probably be good. 12 00:00:43,440 --> 00:00:47,360 Ejaaz: This is a huge ton of events. They've signed a big deal where Anthropic is going 13 00:00:47,360 --> 00:00:52,500 Ejaaz: to buy $5 billion worth of SpaceX's GPUs, also known as Colossus One, 14 00:00:52,500 --> 00:00:54,840 Ejaaz: their first major data center. 15 00:00:55,100 --> 00:00:58,680 Ejaaz: And in turn, they're going to double everyone's Claude restriction limits. 16 00:00:58,920 --> 00:01:01,460 Ejaaz: And Claude is finally going to catch up with OpenAI. 17 00:01:01,860 --> 00:01:04,900 Ejaaz: I am shocked that this partnership is happening, but it's great to see. 18 00:01:05,240 --> 00:01:08,560 Josh: You got to love a CEO that changes his mind. The guy who hated Anthropic, 19 00:01:08,780 --> 00:01:11,220 Josh: he spent a week with them and he was like, you know what? These guys aren't that bad. 20 00:01:11,420 --> 00:01:16,020 Josh: What did he say here? He says, so long as they engage in critical self-examination, 21 00:01:16,200 --> 00:01:17,680 Josh: Claude will probably be good. 22 00:01:17,820 --> 00:01:21,520 Josh: He said, everyone I met was highly competent and cared a great deal about doing the right thing. 23 00:01:21,680 --> 00:01:25,900 Josh: No one set off my evil detector. So he went, he did some background checks and 24 00:01:25,900 --> 00:01:30,240 Josh: he said, yeah, we'll give you the entire Colossus One data center. I'm now your landlord. 25 00:01:30,580 --> 00:01:33,740 Josh: That's a really funny turn of events. I think it's important to highlight the 26 00:01:33,740 --> 00:01:36,760 Josh: context of why this is so important for anyone who has been following along. 27 00:01:37,260 --> 00:01:40,340 Josh: Anthropocos had a really difficult time keeping up with their compute demands, 28 00:01:40,500 --> 00:01:42,540 Josh: mostly because they have grown so quickly. 29 00:01:42,700 --> 00:01:46,660 Josh: In fact, yesterday at Claude's coding event, we had a clip from Dario who was 30 00:01:46,660 --> 00:01:48,380 Josh: kind of doing just a general state of the union. 31 00:01:48,540 --> 00:01:53,360 Josh: And he shared with the public that anthropic was off by a factor of eight in 32 00:01:53,360 --> 00:01:56,280 Josh: projecting their growth. So over the course of the year, they projected 33 00:01:56,960 --> 00:01:59,320 Josh: For about a 10x multiple on the amount of compute they would need. 34 00:01:59,580 --> 00:02:05,520 Josh: The reality is that this year alone, they are now guiding for 80 times what they were expecting. 35 00:02:05,740 --> 00:02:09,040 Josh: A full 8x multiple higher than what they were projecting. 36 00:02:09,280 --> 00:02:12,040 Josh: So therefore, they had a big problem. They couldn't serve the compute. 37 00:02:12,220 --> 00:02:14,740 Josh: The quality of Opus was getting much worse. They needed compute. 38 00:02:15,040 --> 00:02:20,860 Josh: Where is the only large coherent cluster that's sitting dormant? Well, it's owned by XAI. 39 00:02:21,040 --> 00:02:25,360 Josh: It's owned by the SpaceX team, and they signed a deal to give them the entire thing. 40 00:02:25,540 --> 00:02:30,600 Josh: So the whole Colossus 1 cluster that XAI was using to train their early models is now run by Anthropic. 41 00:02:30,880 --> 00:02:33,220 Josh: And as a result, like you mentioned, DJAS, they doubled the rate limits. 42 00:02:33,320 --> 00:02:38,080 Josh: And now Colossus 2, which is the new largest cluster, that's where SpaceX AI 43 00:02:38,080 --> 00:02:40,300 Josh: is going to be training their new data centers. 44 00:02:40,380 --> 00:02:44,140 Josh: Now you'll notice I said SpaceX AI and not XAI, because as of yesterday, 45 00:02:44,300 --> 00:02:46,800 Josh: XAI has been depreciated. It no longer exists. 46 00:02:47,180 --> 00:02:51,680 Josh: SpaceX is kind of rolling everything up into one. And now they are working together 47 00:02:51,680 --> 00:02:56,360 Josh: right down the street anthropic is serving ai from one data center and spacex 48 00:02:56,360 --> 00:03:00,640 Josh: ai is serving data from the other and this is a pretty big change of heart in 49 00:03:00,640 --> 00:03:01,900 Josh: the ai race this is a big deal 50 00:03:01,900 --> 00:03:07,580 Ejaaz: There's so many stories wrapped into this one story both positive and perplexing 51 00:03:07,580 --> 00:03:12,220 Ejaaz: to me that like i don't know i'm just gonna uh mind dump on you for a bit josh 52 00:03:12,220 --> 00:03:15,160 Ejaaz: and like let me know where where you think i end up okay so 53 00:03:15,550 --> 00:03:21,170 Ejaaz: Number one, SpaceX has been, or XAI, has been trailing behind the frontier labs, 54 00:03:21,250 --> 00:03:22,510 Ejaaz: right? They didn't have enough compute. 55 00:03:22,710 --> 00:03:28,110 Ejaaz: We haven't had a good pioneering grok release for like ages at this point, well over a year. 56 00:03:28,430 --> 00:03:32,730 Ejaaz: And so the idea was, okay, if you amass enough GPUs, you can create a better 57 00:03:32,730 --> 00:03:34,450 Ejaaz: model. But that hasn't been the case. 58 00:03:34,650 --> 00:03:40,270 Ejaaz: And an article broke last week that said that XAI hasn't even been utilizing 59 00:03:40,270 --> 00:03:46,230 Ejaaz: all their GPUs. In fact, 11% of the million GPUs that they have hasn't being used for Grok. 60 00:03:46,410 --> 00:03:51,330 Ejaaz: And the idea here would be that Grok just isn't as in demand as some of these 61 00:03:51,330 --> 00:03:52,990 Ejaaz: other models like Claude and ChatGPT. 62 00:03:53,390 --> 00:03:59,090 Ejaaz: So now we see Elon pivoting to a NeoCloud to become an AI compute infrastructure provider. 63 00:03:59,270 --> 00:04:02,410 Ejaaz: And this isn't the first time that he's done it. Literally two weeks ago, 64 00:04:02,530 --> 00:04:05,910 Ejaaz: he signed a deal with Cursor to do the exact same thing. 65 00:04:05,990 --> 00:04:09,550 Ejaaz: And that, by the way, was for Colossus 2. So although we're saying he's using 66 00:04:09,550 --> 00:04:13,410 Ejaaz: Colossus 2 to train Grok models, he's also sectioning off a portion of that 67 00:04:13,410 --> 00:04:16,550 Ejaaz: to allow Cursor to build their own coding model. 68 00:04:16,690 --> 00:04:19,550 Ejaaz: Now, that's a bit of a different deal because the agreement that they've signed 69 00:04:19,550 --> 00:04:23,470 Ejaaz: is if Cursor builds a really good coding model, hey, could you like lend that 70 00:04:23,470 --> 00:04:26,990 Ejaaz: to XAI for Grok as well so we can pass it off as a code code competitor? 71 00:04:27,190 --> 00:04:30,110 Ejaaz: So that's a little more symbiotic between the two companies. 72 00:04:30,130 --> 00:04:34,610 Ejaaz: But for Anthropic, it's pretty clear. Anthropic pays $5 billion for 73 00:04:34,790 --> 00:04:39,030 Ejaaz: Elon gets to register that as revenue so that, you know, XAI makes a bunch of money. 74 00:04:39,170 --> 00:04:43,870 Ejaaz: And for Anthropic, they get access to 300 megawatts of power this month. 75 00:04:44,090 --> 00:04:48,390 Ejaaz: That is a crazy amount of power to have online. And it's pretty clear from Dario's 76 00:04:48,390 --> 00:04:54,090 Ejaaz: words in that clip that we just showed, which is that Anthropic has been lagging behind on compute. 77 00:04:54,230 --> 00:04:56,910 Ejaaz: We finally have the answer as to why we don't have clawed midos. 78 00:04:57,070 --> 00:04:58,990 Ejaaz: In my opinion, it's not because it was too dangerous. 79 00:04:59,190 --> 00:05:01,670 Ejaaz: It's because they didn't have enough compute to serve it in the first place. 80 00:05:01,970 --> 00:05:06,530 Ejaaz: And Living Proof is literally their biggest rival, OpenAI, which delivered 5.5 to everyone. 81 00:05:06,610 --> 00:05:09,970 Ejaaz: And it is called Mythos Little. So the story writes itself. 82 00:05:10,470 --> 00:05:13,950 Josh: Yeah, now Anthropic, they're basically serving compute from every single hyperscaler. 83 00:05:14,010 --> 00:05:17,470 Josh: They have a deal with AWS, with Google, with Microsoft, with NVIDIA, 84 00:05:17,590 --> 00:05:18,910 Josh: with FluidStack, and now SpaceX. 85 00:05:19,150 --> 00:05:21,890 Josh: And SpaceX is the most exciting, the most noteworthy of these, 86 00:05:22,010 --> 00:05:24,550 Josh: because this seems like a pretty large pivot in their business model. 87 00:05:24,670 --> 00:05:31,530 Josh: Now they have Colossus 2, which is currently 500,000 coherent Blackwell chips in a single cluster. 88 00:05:31,630 --> 00:05:34,470 Josh: That's the most powerful in the world so hopefully we start 89 00:05:34,470 --> 00:05:37,270 Josh: to see some amazing stuff coming out of this SpaceX AI team 90 00:05:37,270 --> 00:05:40,090 Josh: but more importantly when you look at the SpaceX business model 91 00:05:40,090 --> 00:05:43,450 Josh: that's performed so well it's very much a services industry they're taking 92 00:05:43,450 --> 00:05:46,510 Josh: satellites and they're blasting them into outer space and they have their own 93 00:05:46,510 --> 00:05:49,490 Josh: business built on top of it with Starlink but they are that service provider 94 00:05:49,490 --> 00:05:53,290 Josh: and owning the rail has proved to be very lucrative as a business model now 95 00:05:53,290 --> 00:05:57,530 Josh: what SpaceX AI is doing currently is kind of the same thing but early stages 96 00:05:57,530 --> 00:06:00,870 Josh: where they have this infrastructure they're incredible at building data centers 97 00:06:00,870 --> 00:06:01,750 Josh: faster than everyone else. 98 00:06:01,870 --> 00:06:04,170 Josh: And now they're leasing it out to other people who need it. 99 00:06:04,330 --> 00:06:07,310 Josh: This applies further out with space as well. 100 00:06:07,410 --> 00:06:12,550 Josh: Now, in the announcement that they spoke about, they very clearly mentioned space AI data centers. 101 00:06:12,890 --> 00:06:15,610 Josh: Anthropic is very excited about this. And it seems as if they're going to be 102 00:06:15,610 --> 00:06:18,570 Josh: working together as well for the space part of the deal. 103 00:06:18,690 --> 00:06:22,110 Josh: So once that Starship 3 rocket comes online this year, once they start getting 104 00:06:22,110 --> 00:06:25,170 Josh: payload into space and they start launching satellites, Anthropic is going to 105 00:06:25,170 --> 00:06:29,230 Josh: have data centers on those rockets into outer space to help with their compute needs. 106 00:06:29,390 --> 00:06:32,890 Josh: So not only is SpaceX AI currently sitting on the largest coherent cluster of 107 00:06:32,890 --> 00:06:34,590 Josh: GPUs with Colossus 2, but 108 00:06:35,080 --> 00:06:39,440 Josh: They now own all of the rails and the only way of getting compute into outer 109 00:06:39,440 --> 00:06:43,340 Josh: space for themselves, but also for every other company in the world who's going 110 00:06:43,340 --> 00:06:45,820 Josh: to need more data centers that are going to need to go out in outer space. 111 00:06:45,960 --> 00:06:47,100 Josh: It's a pretty amazing business they have. 112 00:06:47,260 --> 00:06:50,320 Ejaaz: I think this is a great move for Elon on Xair. 113 00:06:50,420 --> 00:06:54,420 Ejaaz: And I'm in fact more bullish on Xair after they've made this announcement because 114 00:06:54,420 --> 00:06:56,960 Ejaaz: Elon is just doing what he does best. 115 00:06:57,200 --> 00:07:02,280 Ejaaz: He does a zero to one innovation and then he scales the crap out of that. 116 00:07:02,360 --> 00:07:05,540 Ejaaz: He did that with electric cars. He's doing that with humanoid robots, 117 00:07:05,540 --> 00:07:08,300 Ejaaz: and now he's going to do that, hopefully, with compute. 118 00:07:08,480 --> 00:07:13,880 Ejaaz: No one has built a data center as large and as powerful as he has in the time that he has. 119 00:07:14,000 --> 00:07:18,200 Ejaaz: Do you know what I mean? So he just has a very impressive knack for scaling things. 120 00:07:18,540 --> 00:07:22,260 Ejaaz: I want to share this tweet, which I think summarizes this entire story with 121 00:07:22,260 --> 00:07:29,380 Ejaaz: a nice bow tie, which is, XAI is a middling frontier lab and an incredible world-class infra provider. 122 00:07:29,380 --> 00:07:33,820 Ejaaz: Anthropic is an incredible frontier lab and a middling infra provider. 123 00:07:34,020 --> 00:07:40,380 Ejaaz: They had to join forces in order to defeat OpenAI, which is both a good frontier lab with good infra. 124 00:07:40,520 --> 00:07:44,160 Ejaaz: And that's very much how I see this playing out. Anthropic needed compute. 125 00:07:44,660 --> 00:07:48,120 Ejaaz: XAI has a lot of compute needed to sell it, so it works vice versa. 126 00:07:48,840 --> 00:07:52,320 Ejaaz: It's important to note that having Anthropic as one of your major customers 127 00:07:52,320 --> 00:07:55,320 Ejaaz: for compute is probably one of the biggest compliments you can get. 128 00:07:55,900 --> 00:08:00,500 Ejaaz: I just want to point out that Amazon, Google, and I believe Microsoft's 50% 129 00:08:00,500 --> 00:08:04,160 Ejaaz: of their compute backlog is purchased by Anthropic themselves. 130 00:08:04,340 --> 00:08:07,420 Ejaaz: So Anthropic is trying to get their hands on all the compute that makes sense. 131 00:08:07,680 --> 00:08:10,600 Ejaaz: And this is just another partnership that enables that. So cool to see. 132 00:08:10,960 --> 00:08:13,860 Josh: Yeah, they're trying to play catch up and it seems as if they have this. 133 00:08:14,000 --> 00:08:16,460 Josh: It's funny. This was literally the only way they could have got it. 134 00:08:16,460 --> 00:08:20,380 Josh: This is the only compute cluster of this size available on earth for them. 135 00:08:20,640 --> 00:08:22,020 Josh: And they somehow managed to secure it. 136 00:08:22,020 --> 00:08:26,820 Ejaaz: Well, actually, Josh, there is another company whose stock has been underperforming 137 00:08:26,820 --> 00:08:32,160 Ejaaz: for the last month, who also has the second biggest cluster or data center, 138 00:08:32,320 --> 00:08:36,240 Ejaaz: and might also do the same thing because their AI products aren't in demand. 139 00:08:36,580 --> 00:08:39,740 Josh: You guess? Oh, can I take a guess? Yeah. 140 00:08:39,860 --> 00:08:42,920 Josh: Speaking of products that are not in demand, we have meta who has 141 00:08:42,920 --> 00:08:47,800 Josh: also spent a tremendous amount of money on ai and compute and has yes nothing 142 00:08:47,800 --> 00:08:51,420 Josh: to show for it or not much at least so is that who comes next is this the new 143 00:08:51,420 --> 00:08:56,160 Josh: type of business model did like spacex just start the new thing like hey if 144 00:08:56,160 --> 00:09:00,120 Josh: you suck at ai perhaps you can just like use our compute until we figure it 145 00:09:00,120 --> 00:09:02,360 Josh: out and you pay us a lot of money for that i 146 00:09:02,360 --> 00:09:06,640 Ejaaz: I can do you one better um meta actually announced on their previous uh earnings 147 00:09:06,640 --> 00:09:09,700 Ejaaz: uh zuck in a note to to the public said, 148 00:09:10,200 --> 00:09:14,020 Ejaaz: we're making this massive bet because we can't afford to lose the compute race. 149 00:09:14,200 --> 00:09:19,220 Ejaaz: But in the event that our AI products aren't in as demand as we project, 150 00:09:19,660 --> 00:09:23,200 Ejaaz: we can just resell that compute to all the other AI labs that need it. 151 00:09:23,280 --> 00:09:27,080 Ejaaz: So he's preempted what Elon and Anthropic announced yesterday. 152 00:09:27,300 --> 00:09:30,420 Ejaaz: So I think Meta's next and it might actually be bullish for the stock, to be honest. 153 00:09:30,680 --> 00:09:33,480 Josh: And this gets back to our bubble episode from yesterday, which if you haven't 154 00:09:33,480 --> 00:09:34,600 Josh: watched, I highly advise. 155 00:09:34,880 --> 00:09:38,420 Josh: But basically, so long as the demand is there for tokens and people are willing 156 00:09:38,420 --> 00:09:40,020 Josh: to pay for that, everything works. 157 00:09:40,200 --> 00:09:43,740 Josh: So as long as we can continue to build valuable tools, people will figure it out. 158 00:09:43,900 --> 00:09:47,720 Josh: The tokens will get allocated and things will be good. Now, what I found interesting, 159 00:09:47,940 --> 00:09:51,120 Josh: most interesting about this Elon post is the fact that he said he spent last 160 00:09:51,120 --> 00:09:54,240 Josh: week with the Anthropic team because he had a very busy week, 161 00:09:54,400 --> 00:09:56,140 Josh: clearly, because he was doing other things as well, 162 00:09:56,460 --> 00:09:59,600 Josh: most notably the trial that he's currently involved in. 163 00:09:59,700 --> 00:10:03,360 Josh: So he has been going to trial with Sam Altman, 164 00:10:03,780 --> 00:10:06,900 Josh: OpenAI against each other, debating whether or not it is 165 00:10:06,900 --> 00:10:09,840 Josh: okay for open ai to exist as a for-profit entity 166 00:10:09,840 --> 00:10:13,200 Josh: when it started off as a non-profit now thankfully as 167 00:10:13,200 --> 00:10:15,960 Josh: we always do with these court cases we get some good banter we get some 168 00:10:15,960 --> 00:10:18,660 Josh: good clips and we get some good intel in this case 169 00:10:18,660 --> 00:10:21,560 Josh: we have texts from mira marati which for those 170 00:10:21,560 --> 00:10:24,340 Josh: of you who are not familiar she is one of the co-founders of open ai who 171 00:10:24,340 --> 00:10:27,260 Josh: has since left to start their own uh business called 172 00:10:27,260 --> 00:10:32,260 Josh: thinking machines but we have these texts that happened in real time as sam 173 00:10:32,260 --> 00:10:35,980 Josh: altman was getting fired from open ai if you'll remember this was what two years 174 00:10:35,980 --> 00:10:39,580 Josh: ago now uh sam was actually fired as ceo of open ai and completely kicked off 175 00:10:39,580 --> 00:10:43,840 Josh: the company shortly after they brought him back but these texts i read were 176 00:10:43,840 --> 00:10:45,660 Josh: so funny because it gives you an inside look 177 00:10:46,390 --> 00:10:49,850 Josh: Exactly what was happening as the news was going down. So Sam Altman texts Mira. 178 00:10:50,170 --> 00:10:52,630 Josh: He says, can you indicate directionally good or bad? 179 00:10:52,970 --> 00:10:56,190 Josh: Satya and others are anxious as in Satya Nadella, CEO of Microsoft, 180 00:10:56,370 --> 00:10:58,190 Josh: who is the largest stakeholder of OpenAI. 181 00:10:58,510 --> 00:11:03,330 Josh: Mira replies, directionally, very bad. Sam goes, okay, can you wrap up soon? 182 00:11:03,430 --> 00:11:04,970 Josh: Lots of pressure from Microsoft for an update. 183 00:11:05,130 --> 00:11:07,790 Josh: And then Mira just goes, Sam, this is very bad. 184 00:11:08,530 --> 00:11:11,350 Josh: And this is a series of a lot. Sam's like, can I come in? 185 00:11:11,590 --> 00:11:14,790 Josh: And then Mira says, they don't want you to. And then Sam says, what 186 00:11:14,790 --> 00:11:17,670 Josh: do you want to make it better i'm still willing to just walk 187 00:11:17,670 --> 00:11:20,370 Josh: away if that helps if they are ramped up for crazy lawsuits against me 188 00:11:20,370 --> 00:11:23,730 Josh: then i'm not sure what it's just so funny to see into his mind as 189 00:11:23,730 --> 00:11:28,070 Josh: this is going on and you kind of just see the chain of thought of a ceo it's 190 00:11:28,070 --> 00:11:31,210 Josh: really funny he asked like can i come in and talk about a path forward with 191 00:11:31,210 --> 00:11:35,210 Josh: them they're saying no they need more time and it's it's really funny and then 192 00:11:35,210 --> 00:11:38,610 Josh: the one line that has turned into a meme is mira mentioning the new ceo who 193 00:11:38,610 --> 00:11:41,710 Josh: they put in place and the new ceo was um she says 194 00:11:42,420 --> 00:11:47,160 Josh: This new guy is a random Twitch guy. And it turns out that random Twitch guy 195 00:11:47,160 --> 00:11:52,740 Josh: was Emmett Shear, who is the guy who turned Twitch into a billion dollar exit and very successful. 196 00:11:52,800 --> 00:11:55,920 Josh: And he turned it into his ex-header profile background. 197 00:11:56,080 --> 00:11:59,720 Josh: And it was just, it's a very funny story seeing the inside look at what the 198 00:11:59,720 --> 00:12:02,780 Josh: communications looks like during a time as chaotic as this one. 199 00:12:03,440 --> 00:12:07,860 Ejaaz: They're just all human. That's the best part about these lawsuits is we just, 200 00:12:08,060 --> 00:12:11,480 Ejaaz: the public gets to gossip on all of their texts, which just get revealed. 201 00:12:12,240 --> 00:12:17,860 Ejaaz: Now, the other personal text that was revealed, especially over this week, 202 00:12:18,080 --> 00:12:19,580 Ejaaz: was Greg Brockman's diary. 203 00:12:19,700 --> 00:12:22,720 Ejaaz: Now, we referenced this in a previous episode that we did literally last week 204 00:12:22,720 --> 00:12:26,400 Ejaaz: covering the first couple of days of the trial. A lot has happened since. 205 00:12:26,780 --> 00:12:30,480 Ejaaz: A few things, a few highlights that I want to point out. Number one, 206 00:12:30,600 --> 00:12:36,480 Ejaaz: this lady that you're looking at on the screen right now, her name is Siobhan Zilas. 207 00:12:36,580 --> 00:12:39,180 Ejaaz: And she is a partner to Elon. 208 00:12:39,320 --> 00:12:41,940 Ejaaz: They've had like four kids together, But she's also a genius. 209 00:12:42,160 --> 00:12:43,580 Ejaaz: She works at Neuralink, I believe. 210 00:12:43,940 --> 00:12:49,240 Ejaaz: And she was brought on as a witness by the Musk team to basically testify that, 211 00:12:49,400 --> 00:12:52,740 Ejaaz: you know, in the early days, because she was involved in the whole OpenAI founding 212 00:12:52,740 --> 00:12:55,500 Ejaaz: and she did a bunch of work for Elon at that point. 213 00:12:55,900 --> 00:12:59,920 Ejaaz: She wanted to point out that, like, you know, they being Sam and Greg Brockman 214 00:12:59,920 --> 00:13:06,360 Ejaaz: had issues or intentions to make it a full profit eventually and to testify against Sam's nature. 215 00:13:06,360 --> 00:13:11,520 Ejaaz: She was brought to the witness stand, as well as a host of other Musk team witness candidates. 216 00:13:11,760 --> 00:13:16,740 Ejaaz: Now, the other revealing thing was Greg Brockman's diary was exposed, 217 00:13:16,740 --> 00:13:22,020 Ejaaz: and he explicitly says in that diary or in his list of exchanges with Sam that 218 00:13:22,020 --> 00:13:26,980 Ejaaz: we should kind of be strategic about when we turn OpenAI into a for-profit. 219 00:13:27,160 --> 00:13:30,720 Ejaaz: I don't want to do it whilst Elon is involved or is a stakeholder. 220 00:13:30,920 --> 00:13:34,720 Ejaaz: Otherwise, it would seem like we are ousting him. So literally the case that 221 00:13:34,720 --> 00:13:38,660 Ejaaz: Elon Musk's team is making is being proven in the words that Greg is saying. 222 00:13:38,760 --> 00:13:41,320 Ejaaz: Now, his response to that in court was... 223 00:13:41,500 --> 00:13:45,620 Ejaaz: I was just referring to it being when it turns into a billion dollars. 224 00:13:46,120 --> 00:13:52,360 Ejaaz: Not my $29 billion, sorry, not my $30 billion stake that I currently own in open air right now. 225 00:13:52,460 --> 00:13:56,240 Ejaaz: And Musk's team is basically saying, you own $30 billion worth of open air. 226 00:13:56,360 --> 00:14:00,540 Ejaaz: How is this not a full profit? Like, how are you not somewhat incentivized to 227 00:14:00,540 --> 00:14:03,280 Ejaaz: cash this out and make more money or turn it into a full profit? 228 00:14:03,360 --> 00:14:06,120 Ejaaz: And his response was, I was just talking about making a billion dollars, 229 00:14:06,280 --> 00:14:10,440 Ejaaz: not the extra, you know, $29 billion to which Elon's team responded saying, 230 00:14:10,600 --> 00:14:13,040 Ejaaz: why don't you donate the $29 billion then? 231 00:14:13,220 --> 00:14:16,160 Ejaaz: Like, and we can settle this lawsuit up. And so there's a lot of like, 232 00:14:16,540 --> 00:14:18,220 Ejaaz: back and forth going on between them. 233 00:14:18,340 --> 00:14:22,360 Ejaaz: Now, it's been very Elon centric in the news for this court case in particular, 234 00:14:22,360 --> 00:14:25,860 Ejaaz: because Elon's been the first to step up and produce his witnesses. 235 00:14:26,020 --> 00:14:30,060 Ejaaz: But next week is when the script officially flips and Sam takes the witness 236 00:14:30,060 --> 00:14:32,000 Ejaaz: stand as well as OpenAI's witnesses. 237 00:14:32,160 --> 00:14:35,160 Ejaaz: And we might see a similar case where they end up shooting themselves in the 238 00:14:35,160 --> 00:14:38,140 Ejaaz: foot similarly to how Elon has happened for the first week of the trial. 239 00:14:38,660 --> 00:14:41,620 Josh: And what I'm really enjoying is how now they're available via audio. 240 00:14:41,620 --> 00:14:43,080 Josh: So you can actually tune in and listen. 241 00:14:43,260 --> 00:14:46,500 Josh: You can't watch the presentations, but you can listen to the testimonies. 242 00:14:46,600 --> 00:14:49,840 Josh: People have been doing it and kind of sharing the hot takes and the information from it. 243 00:14:50,100 --> 00:14:54,020 Josh: It's very invasive in a way, but in the sense that you just mentioned earlier, 244 00:14:54,160 --> 00:14:57,360 Josh: where people are just human, like obviously Greg Brockman wants to make a billion 245 00:14:57,360 --> 00:15:00,280 Josh: dollars and sure you want the nonprofit, but he's just a human. 246 00:15:00,400 --> 00:15:03,360 Josh: Why would you not? And it's interesting to see these dynamics play out, 247 00:15:03,500 --> 00:15:04,260 Josh: particularly with Siobhan. 248 00:15:04,460 --> 00:15:07,280 Josh: Siobhan is such a sweetheart. She, for people who don't know, she actually 249 00:15:07,280 --> 00:15:10,040 Josh: joined open ai in 2016 which i think was when 250 00:15:10,040 --> 00:15:13,260 Josh: it first got started as a general advisor later moved 251 00:15:13,260 --> 00:15:16,000 Josh: on to the board of directors then went over to neural 252 00:15:16,000 --> 00:15:18,840 Josh: link she was at tesla for a small stint of time she is 253 00:15:18,840 --> 00:15:22,480 Josh: the mother to a few of elon's children there's this whole lore and background 254 00:15:22,480 --> 00:15:26,460 Josh: she actually left the open ai board of directors when elon began xai because 255 00:15:26,460 --> 00:15:30,340 Josh: the conflict of interest so she has become this kind of key feature in this 256 00:15:30,340 --> 00:15:32,940 Josh: early part and it's been interesting to see it play out we're going to continue 257 00:15:32,940 --> 00:15:36,880 Josh: to follow along next week we'll hopefully have some updates It's from the OpenAI side. 258 00:15:37,000 --> 00:15:40,040 Josh: And then the following week, I believe, we're getting close to a verdict. 259 00:15:40,200 --> 00:15:41,400 Josh: So we're going to start to see 260 00:15:41,400 --> 00:15:44,080 Josh: things move quickly over the next few weeks as it relates to the trial. 261 00:15:44,360 --> 00:15:48,200 Ejaaz: Shifting gears slightly, Anthropic, aside from the partnership that they signed, 262 00:15:48,360 --> 00:15:52,960 Ejaaz: has had a pretty promising week of developments, product development specifically, 263 00:15:53,380 --> 00:15:55,780 Ejaaz: as is normal for Anthropic style. 264 00:15:56,800 --> 00:16:01,220 Ejaaz: Finance, they released a big finance update where basically they are creating a 265 00:16:01,460 --> 00:16:06,660 Ejaaz: a bunch of ready-to-run agent templates for the most time-consuming work in financial services. 266 00:16:06,860 --> 00:16:11,580 Ejaaz: Now, if any of you listeners have worked or are working in the financial industry, 267 00:16:11,740 --> 00:16:14,380 Ejaaz: you'll know that there's a bunch of admin work. 268 00:16:14,440 --> 00:16:18,060 Ejaaz: There's a bunch of memos you need to review, a bunch of financial modeling that 269 00:16:18,060 --> 00:16:22,000 Ejaaz: you need to do, a bunch of paperwork and administration that just takes up a 270 00:16:22,000 --> 00:16:26,340 Ejaaz: bunch of time and, in an ideal sense, could be automated, but still requires 271 00:16:26,340 --> 00:16:28,340 Ejaaz: a lot of human input and judgment. 272 00:16:28,640 --> 00:16:34,520 Ejaaz: Well, Claude, specifically Claude Co-Work and Claude Code has now become smart 273 00:16:34,520 --> 00:16:37,760 Ejaaz: enough to be able to assess and do this work for you. 274 00:16:37,980 --> 00:16:42,120 Ejaaz: And they've released a bunch of templates, specifically AI agent workflows that 275 00:16:42,120 --> 00:16:45,360 Ejaaz: can do this for you. Now, it sounds kind of unsexy, right? 276 00:16:45,500 --> 00:16:49,020 Ejaaz: Claude Code and Claude specifically has been able to run Microsoft Excel, 277 00:16:49,220 --> 00:16:51,000 Ejaaz: PowerPoint, and Word pretty well now. 278 00:16:51,080 --> 00:16:54,080 Ejaaz: In fact, they've integrated directly into Microsoft Suite for all of those software 279 00:16:54,080 --> 00:16:56,900 Ejaaz: programs, but there hasn't been a combined package. 280 00:16:57,180 --> 00:16:59,580 Ejaaz: Usually when you get access to these things, you can be like, 281 00:16:59,680 --> 00:17:02,280 Ejaaz: okay, this is cool, but I don't know what to prompt it. What do I say to this? 282 00:17:02,360 --> 00:17:04,180 Ejaaz: How do I connect it to my database and my company? 283 00:17:04,320 --> 00:17:08,200 Ejaaz: How do I make it do the thing that I want it to do? These templates basically solve that. 284 00:17:08,340 --> 00:17:13,200 Ejaaz: And it is once again, Anthropic step towards taking over many different sectors 285 00:17:13,200 --> 00:17:15,920 Ejaaz: that are adjacent to what they do. First, they started off with code. 286 00:17:16,080 --> 00:17:18,980 Ejaaz: Now they're focusing on enterprise, specifically financial services. 287 00:17:19,100 --> 00:17:20,340 Ejaaz: Tomorrow, it could be law. 288 00:17:20,680 --> 00:17:25,060 Ejaaz: It's just a continuing trend of Anthropic doing this particular pattern taking over sectors. 289 00:17:25,620 --> 00:17:28,960 Josh: If you're in college right now and you're in finance and your goal is to make 290 00:17:28,960 --> 00:17:32,500 Josh: it to Wall Street, what are the kind of jobs and the skills that you're optimizing 291 00:17:32,500 --> 00:17:33,780 Josh: for, for that entry-level job, right? 292 00:17:33,800 --> 00:17:37,660 Josh: You're building pitches, you're preparing for meetings, you're reviewing earnings, 293 00:17:37,960 --> 00:17:40,700 Josh: you're building out models, you're doing market research, you're kind of doing 294 00:17:40,700 --> 00:17:41,980 Josh: all of these basic tasks. 295 00:17:42,240 --> 00:17:46,380 Josh: Every single one of these has been turned into one of the financial templates. 296 00:17:46,560 --> 00:17:50,420 Josh: And I'm actually reading from the list of things that Claude has automated because 297 00:17:50,420 --> 00:17:54,080 Josh: it sounds like a job description. And I think this is one of the really powerful 298 00:17:54,080 --> 00:17:56,740 Josh: things that when people are gonna see, they're going to say, 299 00:17:56,820 --> 00:17:58,600 Josh: oh my God, it's taking all the entry-level jobs. 300 00:17:58,720 --> 00:18:02,420 Josh: But the reality is now that child who is in school, who is in college trying 301 00:18:02,420 --> 00:18:04,940 Josh: to be working Wall Street, be in finance, 302 00:18:05,280 --> 00:18:08,280 Josh: has these tools at his disposal to go and accelerate progress 303 00:18:08,280 --> 00:18:11,820 Josh: to do the next level of things what are the things that come after you 304 00:18:11,820 --> 00:18:14,720 Josh: do these market research reviews or the valuation reviews or 305 00:18:14,720 --> 00:18:17,780 Josh: whatever it is that's where i think people should spend a lot of their time and instead 306 00:18:17,780 --> 00:18:21,160 Josh: of viewing this as this detriment to the entry-level class 307 00:18:21,160 --> 00:18:24,240 Josh: of wall street it's like think about how more powerful you can be you 308 00:18:24,240 --> 00:18:27,580 Josh: can just use all of these you have your own employees now so so think bigger 309 00:18:27,580 --> 00:18:31,320 Josh: build better and i think this anthropic update empowers people to do that and 310 00:18:31,320 --> 00:18:35,600 Josh: it's it's cool i i want to test it out i want to apply this to our episode that 311 00:18:35,600 --> 00:18:40,120 Josh: we talked about yesterday of the ai stack and have it kind of build out uh models 312 00:18:40,120 --> 00:18:43,140 Josh: and valuations on what i should get an investment 313 00:18:43,140 --> 00:18:45,380 Ejaaz: Model we should see yeah for this. 314 00:18:45,380 --> 00:18:48,940 Josh: Dude it's like we we don't need to be on wall street or pay hundreds of thousands 315 00:18:48,940 --> 00:18:52,620 Josh: of dollars to hire this entry-level job we just have the templates now on cloth 316 00:18:52,620 --> 00:18:56,160 Josh: and you just ask it questions you use their plugins and it will generate like 317 00:18:56,160 --> 00:18:58,960 Josh: pretty proficient research so i think it's it's very empowering i wonder 318 00:18:58,960 --> 00:19:03,280 Ejaaz: If we could hook that up to like a temporary bank account that we can spin up 319 00:19:03,280 --> 00:19:07,980 Ejaaz: that can then just like invest in a bunch of stocks for us. That might be an episode in itself. 320 00:19:08,260 --> 00:19:12,180 Josh: I mean, again, your imagination is the only fixed variable here. 321 00:19:12,260 --> 00:19:14,880 Josh: It's like the only limitation. It could do basically anything you want. 322 00:19:14,980 --> 00:19:19,200 Ejaaz: I like that. Well, in order to pull all of this off, right, let's assume financial 323 00:19:19,200 --> 00:19:22,900 Ejaaz: services goes crazy and AI is applied to pretty much everything. 324 00:19:23,020 --> 00:19:27,200 Ejaaz: There's a trend. tokens are being consumed, Josh. We need more compute. 325 00:19:27,360 --> 00:19:31,660 Ejaaz: And the one company that has all of the demand but not enough compute is, 326 00:19:31,800 --> 00:19:33,420 Ejaaz: like we mentioned earlier, Anthropic. 327 00:19:33,580 --> 00:19:36,540 Ejaaz: But it's had an inverse side effect, which is 328 00:19:36,860 --> 00:19:42,500 Ejaaz: Google is now the most valuable company in the world, more valuable than NVIDIA, right? 329 00:19:42,580 --> 00:19:46,660 Ejaaz: So we have a new king of the throne, I guess. 330 00:19:47,360 --> 00:19:52,580 Ejaaz: Anthropic signed a $200 billion deal with Google to get access to their compute 331 00:19:52,580 --> 00:19:55,380 Ejaaz: via Google Cloud and through their TPUs. 332 00:19:55,440 --> 00:19:58,760 Ejaaz: Now, it's important to point out that this has spread over five years, 333 00:19:58,880 --> 00:20:00,780 Ejaaz: but it is a locked-in deal. 334 00:20:00,940 --> 00:20:06,000 Ejaaz: In fact, it makes up over 50 to actually 60% of Google's revenue backlog for 335 00:20:06,000 --> 00:20:07,260 Ejaaz: the next couple of years. 336 00:20:07,440 --> 00:20:11,640 Ejaaz: Now, what's important to point out here is Anthropic has already been a big 337 00:20:11,640 --> 00:20:14,660 Ejaaz: fan and customer of Google and it works both ways. 338 00:20:14,760 --> 00:20:18,280 Ejaaz: Number one, Anthropic uses a bunch of Google's TPUs, which is their own version 339 00:20:18,280 --> 00:20:22,840 Ejaaz: of GPUs that competes directly with NVIDIA to train and serve Claude and it's 340 00:20:22,840 --> 00:20:25,160 Ejaaz: actually helped Claude improve massively. 341 00:20:25,280 --> 00:20:27,400 Ejaaz: Mythos was trained on Google's TPUs. 342 00:20:27,600 --> 00:20:31,960 Ejaaz: Now, the other side is Google is one of the biggest stakeholders and investors 343 00:20:31,960 --> 00:20:32,980 Ejaaz: in Anthropic themselves. 344 00:20:33,120 --> 00:20:36,740 Ejaaz: In fact, I think they own roughly right now after this deal and after a few 345 00:20:36,740 --> 00:20:41,160 Ejaaz: other deals that happened over the past month, around 15 to 17% of Anthropic. 346 00:20:41,520 --> 00:20:47,440 Ejaaz: Now it's Google and Amazon combined staked, I think it's like 43% in total of Anthropic. 347 00:20:47,660 --> 00:20:51,380 Ejaaz: So it's just insane. Anthropic is selling a lot of their stake to these competitors 348 00:20:51,380 --> 00:20:55,120 Ejaaz: or infrastructure providers, but there is now a symbiotic relationship where 349 00:20:55,120 --> 00:20:59,080 Ejaaz: Google is more inclined to serve up compute to Anthropic. 350 00:20:59,680 --> 00:21:03,540 Ejaaz: The one thing I couldn't get my head around, Josh, I guess is similar to the 351 00:21:03,540 --> 00:21:06,620 Ejaaz: SpaceX AI and Anthropic partnership is. 352 00:21:07,250 --> 00:21:11,210 Ejaaz: If you're working at Google on the Gemini team, if you're Demis Sibis, 353 00:21:11,490 --> 00:21:14,250 Ejaaz: right, and you know you need compute to make Gemini a better model, 354 00:21:14,410 --> 00:21:18,350 Ejaaz: you must be thinking, could you please just give us that compute so we can train 355 00:21:18,350 --> 00:21:20,750 Ejaaz: the number one frontier model instead of Anthropic and Claude? 356 00:21:20,890 --> 00:21:23,890 Ejaaz: It seems like Google is just kind of like giving them over the reins at this point. 357 00:21:24,150 --> 00:21:27,650 Josh: Well, what's the problem? It's like, is Gemini having an issue? 358 00:21:27,810 --> 00:21:31,590 Josh: Are they hitting a wall where they haven't quite figured out where to spend that compute on? 359 00:21:31,710 --> 00:21:35,170 Josh: Because it's clear Gemini doesn't have nearly as much compute constraints 360 00:21:35,170 --> 00:21:38,090 Josh: in terms of serving up inference to users that anthropic does 361 00:21:38,090 --> 00:21:41,310 Josh: so i wonder if this is a hedge against 362 00:21:41,310 --> 00:21:44,210 Josh: deep mind of the google business like in 363 00:21:44,210 --> 00:21:47,190 Josh: terms of if you are the ceo of google and you have all of your chips on 364 00:21:47,190 --> 00:21:50,190 Josh: deep mind are you going to continue to bet 100 of 365 00:21:50,190 --> 00:21:52,990 Josh: your tpus that deep mind is going to be the one 366 00:21:52,990 --> 00:21:56,070 Josh: to build this incredible ai or are you going to start hedging against 367 00:21:56,070 --> 00:21:58,750 Josh: that by offering your compute to others now there's a few reasons why they 368 00:21:58,750 --> 00:22:01,970 Josh: would offer their compute to others as that hedge but also just because they're having 369 00:22:01,970 --> 00:22:04,870 Josh: the xai problem where they have good models but 370 00:22:04,870 --> 00:22:07,890 Josh: they haven't quite matched the users with that compute and 371 00:22:07,890 --> 00:22:10,630 Josh: they they don't really need all those tpus so we 372 00:22:10,630 --> 00:22:13,850 Josh: could speculate for many reasons but i think the idea is that um 373 00:22:13,850 --> 00:22:17,810 Josh: capitalism rocks man all the arbitrage is going down across all these companies 374 00:22:17,810 --> 00:22:21,350 Josh: if you have any dark gpus they get snapped up immediately and people are willing 375 00:22:21,350 --> 00:22:24,990 Josh: to pay a very large market price in order to do so so the market is efficient 376 00:22:24,990 --> 00:22:29,110 Josh: tpus are being bought and sold all over the place same with gpus and we're just 377 00:22:29,110 --> 00:22:31,990 Josh: going to continue generating these tokens at whatever the cost takes 378 00:22:31,990 --> 00:22:36,190 Ejaaz: Oh my god you just reminded me of something. Did you see that crazy startup 379 00:22:36,190 --> 00:22:43,490 Ejaaz: announcement this week, Josh, where they're putting NVIDIA GPUs and attaching it to your home, 380 00:22:43,750 --> 00:22:49,050 Ejaaz: like to the outside wall, so that your home becomes its own siloed data center, 381 00:22:49,270 --> 00:22:52,650 Ejaaz: which then gets interconnected with a bunch of other homes in your neighborhood, 382 00:22:52,990 --> 00:22:57,510 Ejaaz: effectively, like turning a home neighborhood into a data center. Did you see this? 383 00:22:58,050 --> 00:23:00,550 Josh: Uh, no, but that sounds a little freaky. 384 00:23:00,910 --> 00:23:03,210 Ejaaz: It was the craziest suggestion ever. 385 00:23:03,570 --> 00:23:06,850 Ejaaz: And like NVIDIA's investing into this startup, just to point that out. 386 00:23:07,230 --> 00:23:11,870 Ejaaz: But someone pointed out, which was like, if I see this outside of your home, 387 00:23:11,990 --> 00:23:13,150 Ejaaz: I'm probably going to rob you, right? 388 00:23:13,190 --> 00:23:17,710 Ejaaz: It's the biggest advertising to just to say like, hey, I have all this money, I can afford these GPUs. 389 00:23:17,790 --> 00:23:21,150 Ejaaz: I come anyway, like to talk about your capitalist comment, I was like, 390 00:23:21,230 --> 00:23:22,850 Ejaaz: we might be going too far at this point. 391 00:23:22,990 --> 00:23:26,310 Josh: Oh, well, I'll do you one better quickly, which is this is a plan loosely for 392 00:23:26,310 --> 00:23:27,970 Josh: the Tesla network of autonomous vehicles. 393 00:23:27,990 --> 00:23:31,210 Josh: As they roll out AI5, AI6 chips in these cars, they're going to essentially 394 00:23:31,210 --> 00:23:32,810 Josh: have supercomputers that are connected to the internet. 395 00:23:33,130 --> 00:23:35,270 Josh: And what better use of idle driving 396 00:23:35,270 --> 00:23:38,830 Josh: time than to turn them into a mega cluster. So the idea is out there. 397 00:23:39,030 --> 00:23:41,830 Josh: I'm sure someone is going to capitalize it. And I'm very excited for whatever 398 00:23:41,830 --> 00:23:44,250 Josh: the first person who does that and what it looks like. 399 00:23:44,490 --> 00:23:46,990 Josh: Because man, that'd be pretty cool. I mean, you run some GPUs in your house, 400 00:23:47,090 --> 00:23:50,730 Josh: you get paid for it. You maybe get free electricity, whatever it is. I don't know. 401 00:23:50,950 --> 00:23:53,630 Josh: And then another great business model, which seems like it's not going to work, 402 00:23:53,830 --> 00:23:58,530 Josh: is Eleven Labs, who somehow manages to continue to surprise and delight everyone in the market. 403 00:23:58,650 --> 00:24:02,810 Josh: 11 Labs, for those who aren't familiar, you could think of them as the audio AI company. 404 00:24:02,990 --> 00:24:06,690 Josh: They are the flagship AI company when it comes to anything that enters your 405 00:24:06,690 --> 00:24:09,990 Josh: ears, whether it be voices, whether it be music, anything that you hear, 406 00:24:10,350 --> 00:24:12,910 Josh: 11 Labs has specialized in it. And now they're not the only ones. 407 00:24:13,090 --> 00:24:16,550 Josh: Google has an audio model that makes music really well. There's a few other 408 00:24:16,550 --> 00:24:20,430 Josh: companies that are trying, but for some reason, 11 Labs is doing it the best. 409 00:24:20,510 --> 00:24:25,090 Josh: And it's thanks to their CEO, Matty, who had an announcement of $500 million 410 00:24:25,090 --> 00:24:27,670 Josh: of ARR this year. This is a huge announcement from the team. 411 00:24:28,110 --> 00:24:31,690 Ejaaz: Yeah, super impressive. One thing that 11 Labs has done well is they've, 412 00:24:32,330 --> 00:24:37,210 Ejaaz: manage to maintain and build a really cool, attractive image on the outside. 413 00:24:37,350 --> 00:24:42,130 Ejaaz: So many celebrities are partnered or invested in 11 Labs, the main one being Matthew McConaughey. 414 00:24:42,310 --> 00:24:46,570 Ejaaz: We've got Jamie Foxx joining us around, Eva Longoria. So that helps with your 415 00:24:46,570 --> 00:24:51,090 Ejaaz: distribution mode of making 11 Labs cool and appealing to the people that they're 416 00:24:51,090 --> 00:24:53,450 Ejaaz: eventually going to be replicating or replacing, right? 417 00:24:53,530 --> 00:24:56,170 Ejaaz: Like there's a whole fear in Hollywood that it's like, you know, 418 00:24:56,570 --> 00:24:58,690 Ejaaz: AI is going to replace us, it's not authentic. 419 00:24:58,970 --> 00:25:03,490 Ejaaz: But if you can kind of license your IP, your voice, your character in a way 420 00:25:03,490 --> 00:25:07,510 Ejaaz: that you can control, such as with 11 labs, you can end up, you know, 421 00:25:07,610 --> 00:25:10,830 Ejaaz: making a bunch of money and 500 million AR kind of like plays into that. 422 00:25:10,930 --> 00:25:12,670 Ejaaz: So a massively successful model. 423 00:25:12,850 --> 00:25:15,610 Ejaaz: But there is another story that I saw today. 424 00:25:15,910 --> 00:25:20,790 Ejaaz: Well, rather, not a story, Josh, but a clip. Have you ever seen a robot, right? 425 00:25:21,150 --> 00:25:24,490 Ejaaz: Crack an egg before with such precision? I'll do you one better. 426 00:25:24,610 --> 00:25:27,990 Ejaaz: Have you ever seen a robot carefully take a pipette and 427 00:25:27,990 --> 00:25:31,750 Ejaaz: pipette the right amount of stuff into a petri dish or have you ever seen it 428 00:25:31,750 --> 00:25:35,590 Ejaaz: solve a rubik's cube or have you ever seen it what is it doing here is this 429 00:25:35,590 --> 00:25:39,030 Ejaaz: like blending and making a shake anyway the point i'm saying is this new startup 430 00:25:39,030 --> 00:25:43,210 Ejaaz: announcement from genesis ai has created a new generalized robotics model that 431 00:25:43,210 --> 00:25:45,650 Ejaaz: can be applied to any kind of robotic machinery. 432 00:25:46,760 --> 00:25:52,840 Ejaaz: Instantly give it a very precise amount of movement and precision in the way 433 00:25:52,840 --> 00:25:56,840 Ejaaz: that it interacts with different objects to a point that we have never seen before. 434 00:25:57,100 --> 00:26:00,900 Ejaaz: I think robotics is finally reaching escape velocity. We've seen a bunch of 435 00:26:00,900 --> 00:26:04,080 Ejaaz: announcements from the likes of Figure and Unitree, which, you know, 436 00:26:04,140 --> 00:26:07,300 Ejaaz: the robot's doing cool things, but it's mainly at a manufacturing process level, 437 00:26:07,400 --> 00:26:08,320 Ejaaz: like they work in factories. 438 00:26:08,580 --> 00:26:11,500 Ejaaz: This is the first time where I'm like, oh, the robot looks pretty human and 439 00:26:11,500 --> 00:26:13,880 Ejaaz: I might have it in my home. It's just pretty cool. 440 00:26:13,880 --> 00:26:16,640 Josh: Yeah okay so we're going to speed around the rest of these because a few 441 00:26:16,640 --> 00:26:19,380 Josh: interesting news topics that are left one is a new round that 442 00:26:19,380 --> 00:26:23,460 Josh: is led by peter teal for a company named panthalassa 443 00:26:23,460 --> 00:26:26,160 Josh: i think i'm saying that right um 140 million dollar 444 00:26:26,160 --> 00:26:29,000 Josh: round led by peter teal and their entire basis of this 445 00:26:29,000 --> 00:26:33,940 Josh: company is to build planetary scale energy in the middle of the ocean so basically 446 00:26:33,940 --> 00:26:37,540 Josh: they have these you could think of a compute cluster a data center that we have 447 00:26:37,540 --> 00:26:41,080 Josh: kind of like in texas or that we have in tennessee what they're doing is they're 448 00:26:41,080 --> 00:26:44,700 Josh: taking that data center they're making a float and then they're driving it off 449 00:26:44,700 --> 00:26:47,180 Josh: into the middle of the ocean to function independently. 450 00:26:47,460 --> 00:26:50,800 Josh: So you can unlock the ocean as a planetary scale resource and 451 00:26:51,470 --> 00:26:54,930 Josh: This is pretty mind-blowing when we think about how AI is going to scale. 452 00:26:55,210 --> 00:26:58,970 Josh: I mean, we are very familiar now with the AI data centers in space conversation, 453 00:26:58,970 --> 00:27:02,530 Josh: but the AI in the ocean seems pretty noteworthy, pretty interesting. 454 00:27:02,950 --> 00:27:07,570 Ejaaz: And finally, we have some, I guess, AI-adjacent news, but not really. 455 00:27:07,710 --> 00:27:10,770 Ejaaz: We were just blown away by this, and we had to call it out. 456 00:27:10,890 --> 00:27:16,170 Ejaaz: Guys, the next coronavirus apparently is around the corner, and it's appropriately named Hantavirus. 457 00:27:16,330 --> 00:27:20,310 Ejaaz: There's a luxury cruise trip right now where it has an outbreak, 458 00:27:20,310 --> 00:27:22,850 Ejaaz: very reminiscent of coronavirus. I'm getting PTSD. 459 00:27:23,430 --> 00:27:27,570 Ejaaz: And nearly 150 people are still on board with no symptoms, but they are some 460 00:27:27,570 --> 00:27:30,250 Ejaaz: seriously ill passengers. Josh, what's going on here? 461 00:27:30,490 --> 00:27:32,570 Josh: Dude, I woke up this morning. This was my entire feed online. 462 00:27:32,730 --> 00:27:34,170 Josh: I'm like, what is going on? Is the world ending? 463 00:27:34,530 --> 00:27:38,270 Josh: And when we recorded our episode yesterday on the threats to the AI bubble, 464 00:27:38,430 --> 00:27:41,270 Josh: this was not baked into the equation. So we're just sharing this just to bake 465 00:27:41,270 --> 00:27:42,670 Josh: it in. This is highly speculative. 466 00:27:43,150 --> 00:27:46,230 Josh: According to Polymarket, there's only a 9% chance this turns into a pandemic. 467 00:27:46,230 --> 00:27:49,030 Josh: But it is like it's funny and not funny 468 00:27:49,030 --> 00:27:51,910 Josh: but like it is scary to see the kind of rerun of 469 00:27:51,910 --> 00:27:54,930 Josh: this traumatic event that happened but in a different flavor in this different 470 00:27:54,930 --> 00:27:58,030 Josh: context and i just wanted to just highlight as another thing to you know keep 471 00:27:58,030 --> 00:28:01,550 Josh: your eyes on everything's fine for now but people in at least three u.s states 472 00:28:01,550 --> 00:28:05,870 Josh: are reportedly being monitored for possible exposure and like a people have 473 00:28:05,870 --> 00:28:10,190 Josh: died and things are happening so just you know want to make sure everyone's aware everyone's okay 474 00:28:10,750 --> 00:28:13,770 Josh: things are going to be fine. According to Polymarket, the odds are low. 475 00:28:13,930 --> 00:28:14,970 Josh: They're still in our favor. 476 00:28:15,210 --> 00:28:19,110 Josh: The up-only pump is going to continue, but this should be hedged into your models at least. 477 00:28:19,310 --> 00:28:22,050 Josh: Add this to your LLMs. Tell them, hey, create this as a parameter when you're 478 00:28:22,050 --> 00:28:23,430 Josh: evaluating risk going forward. 479 00:28:23,570 --> 00:28:28,570 Ejaaz: Yes, yes. Well, there you have it, folks. You are officially caught up by the 480 00:28:28,570 --> 00:28:31,050 Ejaaz: end of the week of everything that's been happening in AI. 481 00:28:31,250 --> 00:28:35,190 Ejaaz: Everything from the drama in the courtroom to the latest partnerships between 482 00:28:35,190 --> 00:28:40,190 Ejaaz: the rivals that are to kill the other rival, all the way to breaking virus news. 483 00:28:40,430 --> 00:28:42,430 Ejaaz: You can get everything here at Limitless. 484 00:28:42,750 --> 00:28:47,210 Ejaaz: We released a bunch of episodes this week, which jumped into the depths of why 485 00:28:47,210 --> 00:28:49,910 Ejaaz: all the semiconductor stocks are pumping massively. 486 00:28:50,150 --> 00:28:53,970 Ejaaz: We covered OpenAI's new phone that's releasing as early as 2027. 487 00:28:54,190 --> 00:28:58,470 Ejaaz: And we compared the motherlode of all AI models, GPT 5.5 versus Claude, 488 00:28:58,570 --> 00:29:01,830 Ejaaz: Opus 4.7, which one's the best. We did a breakdown. We did a bunch of demos. 489 00:29:02,010 --> 00:29:06,190 Ejaaz: We may or may not have recreated Mario, definitely go check out all those episodes. 490 00:29:06,690 --> 00:29:09,430 Ejaaz: If you are listening to us on whatever platform you're on, YouTube, 491 00:29:09,710 --> 00:29:12,550 Ejaaz: Spotify, or Apple, please leave us a review, give us a rating. 492 00:29:12,710 --> 00:29:13,870 Ejaaz: It helps us out massively. 493 00:29:14,390 --> 00:29:15,910 Ejaaz: Josh, any final parting thoughts? 494 00:29:16,590 --> 00:29:18,890 Josh: Thank you so much for watching. If you made it this far, you're a real one. 495 00:29:19,010 --> 00:29:22,270 Josh: I very much appreciate all the support of getting caught up. Go enjoy your weekend. 496 00:29:22,430 --> 00:29:24,090 Josh: Have a great time touching grass because you're all caught up. 497 00:29:24,210 --> 00:29:26,350 Josh: Thank you so much for watching and we will see you guys next week. 498 00:29:26,470 --> 00:29:26,950 Ejaaz: See you guys.