1 00:00:00,020 --> 00:00:04,020 Josh: If you were alive in 2008, chances are you remember the market crashing, 2 00:00:04,340 --> 00:00:08,120 Josh: a catastrophic failure of the housing market that bled into the entire economy, 3 00:00:08,120 --> 00:00:09,780 Josh: and it wiped out a lot of entities. 4 00:00:09,880 --> 00:00:13,920 Josh: And if you'll remember, there was one person in particular who profited a tremendous 5 00:00:13,920 --> 00:00:17,620 Josh: amount off of this happening, and his name is Michael Burry. He made $100 million. 6 00:00:17,880 --> 00:00:23,140 Josh: Personally, collectively, he made a billion dollars off of shorting this market crash. 7 00:00:23,280 --> 00:00:26,660 Josh: He saw what was right early, and he doubled down on it, and he made a killing. 8 00:00:26,660 --> 00:00:30,700 Josh: In fact, so much so that it became a movie that you've probably seen named The Big Short. 9 00:00:31,080 --> 00:00:35,740 Josh: Now, Michael Berry recently has published a new position in recent SEC filings that we uncovered. 10 00:00:36,020 --> 00:00:39,260 Josh: And Ejaz has been all over this over the last couple of weeks, 11 00:00:39,640 --> 00:00:42,480 Josh: tracking the positions, trying to understand why he's making these things. 12 00:00:42,540 --> 00:00:46,760 Josh: And if this bet is correct in predicting the next big bubble to pop, 13 00:00:46,860 --> 00:00:48,620 Josh: which is seeming to be AI. 14 00:00:48,860 --> 00:00:52,020 Josh: So he predicted the housing bubble in 08. He made a billion dollars. 15 00:00:52,200 --> 00:00:57,840 Josh: He's predicting another bubble in 2025. five, is he going to make another billion dollars, EJS? 16 00:00:58,060 --> 00:01:01,880 Ejaaz: The short answer is, I don't think he will. I think he's going to get blown out. 17 00:01:03,120 --> 00:01:09,440 Ejaaz: So Michael Burry is back. He is short around $300 million worth of Nvidia shares. 18 00:01:09,580 --> 00:01:13,480 Ejaaz: So he's shorting, at this point, the richest, wealthiest company in the world. 19 00:01:13,620 --> 00:01:17,200 Ejaaz: And that is a big sign, basically saying that I think the air bubble is popping. 20 00:01:17,380 --> 00:01:18,980 Ejaaz: So the question is, is it? 21 00:01:19,700 --> 00:01:24,320 Ejaaz: Let's go through his thesis, Josh. I'm going to simplify his post here. 22 00:01:24,440 --> 00:01:28,360 Ejaaz: Which is basically GPUs run the AI world, right? 23 00:01:28,800 --> 00:01:34,400 Ejaaz: Trillions of dollars have been spent by Frontier AI labs collectively to train state-of-the-art AI. 24 00:01:34,940 --> 00:01:38,880 Ejaaz: But the thing with these GPUs, Josh, is that they have a lifespan, 25 00:01:39,180 --> 00:01:41,020 Ejaaz: right? They don't last forever, right? 26 00:01:41,260 --> 00:01:44,620 Ejaaz: And he estimates that the top Frontier labs. 27 00:01:46,180 --> 00:01:50,720 Ejaaz: Overestimating or artificially boosting the lifespan of the GPUs that they purchase. 28 00:01:50,860 --> 00:01:53,560 Ejaaz: And you might be like, well, that sounds boring. Why is that interesting? 29 00:01:53,960 --> 00:01:57,080 Ejaaz: Well, the thing is, if you have assets on your balance sheet, 30 00:01:57,240 --> 00:01:59,180 Ejaaz: that factors into your stock price. 31 00:01:59,360 --> 00:02:03,920 Ejaaz: So the point that he's making is all these big companies are artificially boosting 32 00:02:03,920 --> 00:02:08,700 Ejaaz: the lifespan of the GPUs that they've purchased so that it inflates their stock price. 33 00:02:08,900 --> 00:02:11,840 Ejaaz: And therefore, the real stock price is actually a lot lower. 34 00:02:12,000 --> 00:02:16,200 Ejaaz: In fact, he says at the bottom of this tweet, by 2028, Oracle will overstate 35 00:02:16,200 --> 00:02:19,900 Ejaaz: earnings by 26.9%, Meta by 20%, etc. 36 00:02:20,120 --> 00:02:25,800 Ejaaz: And so his big short, Josh, is on the behemoth that is supplying all these Frontier 37 00:02:25,800 --> 00:02:29,720 Ejaaz: AI labs, NVIDIA, saying that when the bubble pops, and he thinks the bubble's 38 00:02:29,720 --> 00:02:34,240 Ejaaz: going to pop now, he will make a heck of a lot of money on this trade. 39 00:02:34,280 --> 00:02:37,200 Josh: So is he saying the bubble's going to pop now, or is it going to pop in 2028? 40 00:02:37,360 --> 00:02:39,540 Josh: Because he's referencing those numbers in 2028. 41 00:02:39,680 --> 00:02:44,280 Josh: So I'm kind of curious how he's framing this is this like because i'm now i'm 42 00:02:44,280 --> 00:02:47,980 Josh: thinking about our positions um is this something that he predicts is going 43 00:02:47,980 --> 00:02:51,620 Josh: to happen in 2028 or is this more of a short timescale position if 44 00:02:51,620 --> 00:02:55,200 Ejaaz: We remember from his original short i think he held a position for about a year, 45 00:02:56,290 --> 00:03:00,390 Ejaaz: Or at least like the better part of a year. So I think he probably estimates 46 00:03:00,390 --> 00:03:02,890 Ejaaz: that it's going to happen sometime within that time frame. 47 00:03:03,170 --> 00:03:08,210 Ejaaz: But I have more evidence as to like why he might be thinking that the bubble is bursting. 48 00:03:08,730 --> 00:03:11,150 Ejaaz: Have you heard of what a neocloud is, Josh? 49 00:03:12,410 --> 00:03:16,790 Josh: The name is familiar, but I don't understand quite what that means. So please fill me in. 50 00:03:17,050 --> 00:03:21,090 Ejaaz: Okay, so think of like AWS or any of these other typical cloud providers. 51 00:03:21,090 --> 00:03:27,110 Ejaaz: But they're specifically focused on providing GPUs and data center pipelines for you. 52 00:03:27,530 --> 00:03:31,170 Ejaaz: So think of like an AWS just for AI compute. That makes sense? 53 00:03:31,810 --> 00:03:38,670 Ejaaz: And so what he did was he looked at the major NeoCloud providers and he looked 54 00:03:38,670 --> 00:03:43,630 Ejaaz: at their stock prices over the last week, Josh, and they were down. 55 00:03:43,670 --> 00:03:48,670 Ejaaz: So he upped his position and was like, okay, this is the bubble bursting. I'm gonna lay my claim. 56 00:03:49,240 --> 00:03:53,500 Ejaaz: There's an issue with this thesis, Josh. I think he's dead wrong. 57 00:03:53,500 --> 00:03:56,760 Ejaaz: And I'm about to walk you through four reasons why he's wrong. 58 00:03:57,220 --> 00:03:59,540 Josh: Okay. So then just to kind of like wrap my head around this, 59 00:03:59,660 --> 00:04:02,700 Josh: the NeoClouds are kind of infrastructure for people who don't want to build 60 00:04:02,700 --> 00:04:03,500 Josh: their own infrastructure. 61 00:04:03,900 --> 00:04:06,480 Josh: Kind of like you described AWS. If you want to get a data center, 62 00:04:06,600 --> 00:04:09,760 Josh: but you don't want to build a data center, you offload that responsibility to a NeoCloud. 63 00:04:10,020 --> 00:04:14,960 Josh: And he thinks that the cost that these companies are marking down these NeoCloud 64 00:04:14,960 --> 00:04:19,060 Josh: prices at, particularly the GPUs is much higher because the depreciation happens 65 00:04:19,060 --> 00:04:21,740 Josh: on a shorter timescale than the companies are writing off. 66 00:04:22,100 --> 00:04:28,060 Ejaaz: Correct. And NeoClouds is one part of the picture, but it's a great example 67 00:04:28,060 --> 00:04:33,720 Ejaaz: to lead with because so many frontier AI labs like Microsoft OpenAI actually 68 00:04:33,720 --> 00:04:36,240 Ejaaz: pay these NeoClouds billions of dollars. 69 00:04:36,420 --> 00:04:40,320 Ejaaz: Like Microsoft just signed a $19 billion contract with a NeoCloud provider called 70 00:04:40,320 --> 00:04:43,220 Ejaaz: Nebius, which kind of sent its stock price going up. 71 00:04:43,340 --> 00:04:46,240 Ejaaz: We talked about it on a previous episode. so like it's a 72 00:04:46,240 --> 00:04:49,420 Ejaaz: good example to kind of like lead with but the 73 00:04:49,420 --> 00:04:52,360 Ejaaz: take here is that these neoclouds are 74 00:04:52,360 --> 00:04:57,840 Ejaaz: basically as you said overestimating the life cycle of these gpus and therefore 75 00:04:57,840 --> 00:05:02,300 Ejaaz: they are wrong except that's not the case at all and i present to you my first 76 00:05:02,300 --> 00:05:06,500 Ejaaz: counter thesis which is he's wrong about the two to three year depreciation 77 00:05:06,500 --> 00:05:11,460 Ejaaz: cycle it's actually more like five to six in some cases even eight years. 78 00:05:11,580 --> 00:05:15,180 Ejaaz: These GPUs are used for more than just training AI models. 79 00:05:15,380 --> 00:05:19,740 Ejaaz: They are used for inference, and they're sometimes used just for general queries 80 00:05:19,740 --> 00:05:21,900 Ejaaz: and distribution for these AI models, right? 81 00:05:22,320 --> 00:05:25,840 Ejaaz: But the most staggering kind of example comes from the top doc, 82 00:05:26,000 --> 00:05:31,380 Ejaaz: the top neocloud called CoreWeave, who kind of shows the opposite to what Michael 83 00:05:31,380 --> 00:05:36,440 Ejaaz: Burry is estimating, which is, there are people booking up GPUs, Josh. 84 00:05:37,250 --> 00:05:40,270 Ejaaz: Two quarters in advance, six months in advance. 85 00:05:40,430 --> 00:05:46,250 Ejaaz: And these aren't the latest GPUs. These are GPUs that are like five years old, 86 00:05:46,550 --> 00:05:49,210 Ejaaz: three years old, in some cases, even longer. 87 00:05:49,450 --> 00:05:53,710 Ejaaz: So if you dig into the details of all these different NeoClouds, 88 00:05:53,850 --> 00:05:57,510 Ejaaz: you'll start realizing that it's the opposite of what Michael Burry is estimating. 89 00:05:57,890 --> 00:06:01,710 Ejaaz: One, these GPUs have a very long lifecycle. And two, they're being used for 90 00:06:01,710 --> 00:06:05,930 Ejaaz: way longer than people expect for really important things within AI. 91 00:06:06,150 --> 00:06:09,090 Ejaaz: They are oversubscribed, they are overutilized. 92 00:06:09,510 --> 00:06:12,910 Josh: This is interesting because you're getting like these really two counterpoints 93 00:06:12,910 --> 00:06:15,950 Josh: here where Michael Burry's subjective take where he's like, no, this is wrong. 94 00:06:16,110 --> 00:06:18,730 Josh: And then you have this objective number, which is CoreWeave. 95 00:06:18,770 --> 00:06:21,810 Josh: And they're saying, wait a second, we're actually selling out these old hard 96 00:06:21,810 --> 00:06:23,430 Josh: drives or these old GPUs. 97 00:06:24,030 --> 00:06:29,290 Josh: Two quarters in advance. So how do we kind of piece together who's right and 98 00:06:29,290 --> 00:06:30,350 Josh: who's wrong through this? 99 00:06:30,470 --> 00:06:33,730 Josh: Because it seems like the market demand, and intuitively it makes sense too, 100 00:06:33,870 --> 00:06:36,190 Josh: that there is no shortage of people who want to generate tokens. 101 00:06:36,350 --> 00:06:40,330 Josh: And even if you are paying a little extra in premium in terms of cost per kilowatt 102 00:06:40,330 --> 00:06:44,230 Josh: for those older GPUs, it's probably still worthwhile because the amount of money 103 00:06:44,230 --> 00:06:45,990 Josh: you could build on top of that is so large. 104 00:06:46,130 --> 00:06:50,690 Ejaaz: Forget about GPUs. Even the CPUs are being used for all the AI distribution 105 00:06:50,690 --> 00:06:55,750 Ejaaz: and coordination. I mean, look at this, AMD CPU tab has just gone up in this 106 00:06:55,750 --> 00:06:57,850 Ejaaz: week's projected earnings, right? 107 00:06:58,270 --> 00:07:01,950 Ejaaz: But to answer your question, it's like, okay, well, who's right here, right? 108 00:07:02,670 --> 00:07:07,310 Ejaaz: I want to kind of, before I address that point, I want to look at another actor 109 00:07:07,310 --> 00:07:10,030 Ejaaz: within this whole circular economy that we like to talk about, 110 00:07:10,130 --> 00:07:12,390 Ejaaz: right? Which is the Frontier AI Labs, right? 111 00:07:13,010 --> 00:07:17,130 Ejaaz: Google, who is an established company, they just had their first $100 billion 112 00:07:17,130 --> 00:07:22,350 Ejaaz: revenue quarter, are using eight-year-old GPUs to like fund their whole thing. 113 00:07:22,470 --> 00:07:27,730 Ejaaz: So again, another data point saying that I don't think these GPUs are old and or not useful. 114 00:07:28,150 --> 00:07:32,890 Ejaaz: And then if you look at NVIDIA themselves, they also have crazy amounts of demand. 115 00:07:33,010 --> 00:07:37,090 Ejaaz: They're oversubbed for years in advance for their Blackwell GPUs and all their 116 00:07:37,090 --> 00:07:38,610 Ejaaz: new GPUs going forwards. 117 00:07:39,170 --> 00:07:45,070 Ejaaz: So then the last actor that I would want to look at is where does all the demand come from? 118 00:07:45,170 --> 00:07:47,790 Ejaaz: Like who's buying these GPUs? Are they satiating demand? 119 00:07:48,070 --> 00:07:53,270 Ejaaz: And if you look at every other company like Google, who has a ton of end users, 120 00:07:53,490 --> 00:07:55,270 Ejaaz: OpenAI with chat GPT users, et cetera. 121 00:07:55,970 --> 00:08:01,230 Ejaaz: Every insider take points to there's not enough GPUs to supply all of this demand 122 00:08:01,230 --> 00:08:04,310 Ejaaz: that they're getting from the apps that they produce, from Sora, 123 00:08:04,450 --> 00:08:05,670 Ejaaz: from a bunch of these different things. 124 00:08:05,810 --> 00:08:11,150 Ejaaz: There's just too much demand. And in many cases, they don't have enough scale 125 00:08:11,150 --> 00:08:13,370 Ejaaz: to even kind of like meet this demand. 126 00:08:13,910 --> 00:08:17,570 Ejaaz: Jensen Huang had a conversation with TSMC, which is like their main 127 00:08:18,000 --> 00:08:22,000 Ejaaz: chip manufacturer asking them to increase the rates of production by 50%. 128 00:08:22,000 --> 00:08:25,820 Ejaaz: So the point I'm saying is Michael Burry, I think is dead wrong here. 129 00:08:26,040 --> 00:08:29,840 Ejaaz: He's definitely underestimated demand and he's underestimated the life cycle 130 00:08:29,840 --> 00:08:32,980 Ejaaz: of these GPUs. They are more valuable than he thought. 131 00:08:33,160 --> 00:08:37,860 Ejaaz: And the old GPUs are worth more than they were worth when they sold originally. 132 00:08:38,080 --> 00:08:41,560 Ejaaz: They're still selling within 5% of their contracts, which is just crazy. 133 00:08:41,780 --> 00:08:44,860 Josh: So maybe I have to ask a pretty dumb question or seemingly 134 00:08:44,860 --> 00:08:50,400 Josh: dumb at least and it's like why why is that so important to the bubble how companies 135 00:08:50,400 --> 00:08:55,280 Josh: factor off losses in gpus over time is this some real big existential threat 136 00:08:55,280 --> 00:09:00,660 Josh: to me it feels like a small part to a bigger picture and not quite as big as 137 00:09:00,660 --> 00:09:04,600 Josh: something that would totally blow out an entire market that's been built so far the 138 00:09:04,600 --> 00:09:10,780 Ejaaz: Biggest budget spend for any major ai company is on GPUs. 139 00:09:11,200 --> 00:09:14,080 Ejaaz: That's where all the hundreds of billions are going for from Microsoft. 140 00:09:14,480 --> 00:09:19,020 Ejaaz: That's where the majority of OpenAI's $1.4 trillion that they've committed over 141 00:09:19,020 --> 00:09:20,320 Ejaaz: the next five years is going to. 142 00:09:20,460 --> 00:09:23,620 Ejaaz: It's all to these compute providers. It's all to these neoclouds. 143 00:09:23,700 --> 00:09:27,760 Ejaaz: It's all to creating their own chips. It's all to Jensen Huang's pocket. 144 00:09:28,340 --> 00:09:32,360 Ejaaz: And so most of the CapEx bubble comes from GPUs. 145 00:09:32,460 --> 00:09:35,760 Ejaaz: So if the bubble is going to burst anywhere, it's going to be from over-leveraging 146 00:09:35,760 --> 00:09:37,880 Ejaaz: on GPUs. Do you want to know something crazy. 147 00:09:38,260 --> 00:09:38,740 Josh: Josh? What's that? 148 00:09:38,860 --> 00:09:44,960 Ejaaz: Even though hundreds of billions have been spent this year alone on GPUs, 149 00:09:45,200 --> 00:09:50,020 Ejaaz: the companies who have spent it haven't even made a dent on their balance sheet. 150 00:09:50,520 --> 00:09:53,620 Ejaaz: Remember, we're talking about Google here. We're talking about Microsoft here. 151 00:09:53,800 --> 00:09:56,940 Ejaaz: They make hundreds of billions of dollars in net profit per year. 152 00:09:57,100 --> 00:10:02,320 Ejaaz: They have plenty of flush cash and they're spending it on machinery that they think is going, 153 00:10:03,830 --> 00:10:07,870 Ejaaz: you and I maybe don't see on the enterprise side or on the end consumer side, 154 00:10:07,970 --> 00:10:09,610 Ejaaz: but they are obviously seeing it. 155 00:10:10,190 --> 00:10:13,610 Josh: Okay. That makes sense. So the depreciation really is a big factor because it's 156 00:10:13,610 --> 00:10:16,070 Josh: kind of artificially inflating numbers across the board. 157 00:10:16,190 --> 00:10:18,770 Josh: It's from the people who are borrowing it, the people who are issuing the GPUs. 158 00:10:19,010 --> 00:10:21,870 Josh: There's lots of high inflation that he's assuming is being baked into this. 159 00:10:22,050 --> 00:10:25,850 Josh: And eventually that inflation kind of fizzles its way out through either some 160 00:10:25,850 --> 00:10:30,510 Josh: really aggressive event or just over time degrading, which makes sense. 161 00:10:30,630 --> 00:10:34,010 Josh: Okay. I think i think i'm up to speed on this i think i get his case 162 00:10:34,010 --> 00:10:36,830 Josh: but what seems a little bit different this time is that 163 00:10:36,830 --> 00:10:39,550 Josh: the first short that was made was in 2008 it 164 00:10:39,550 --> 00:10:43,070 Josh: was against basically the entire stock market and particularly the housing market um 165 00:10:43,070 --> 00:10:47,270 Josh: that was not growing nearly as fast as the ai market is that's like shorting 166 00:10:47,270 --> 00:10:53,530 Josh: the internet and if you shorted the internet before 1999 or after like there's 167 00:10:53,530 --> 00:10:56,770 Josh: this very narrow window to be correct without getting blown out because what 168 00:10:56,770 --> 00:10:59,190 Josh: a lot of people don't realize is in order to short a company you have to borrow 169 00:10:59,190 --> 00:11:01,410 Josh: money against from somewhere else. 170 00:11:01,550 --> 00:11:04,930 Josh: And that borrowing comes with a premium. You have to pay a certain percentage 171 00:11:04,930 --> 00:11:06,770 Josh: interest in order to loan the money. 172 00:11:06,870 --> 00:11:09,970 Josh: If he's not right on this, or even if he's wrong by a couple of months, 173 00:11:10,130 --> 00:11:14,750 Josh: with the rate that the market is accelerating, it feels like it's a very difficult 174 00:11:14,750 --> 00:11:15,850 Josh: thing to get right because... 175 00:11:16,490 --> 00:11:21,070 Josh: You not only have to time it right, but you can't get blown out by the appreciation 176 00:11:21,070 --> 00:11:22,690 Josh: of things as fast as they're growing. 177 00:11:22,830 --> 00:11:26,570 Josh: Like I've never seen a hockey stick of an industry steeper than this one. 178 00:11:26,710 --> 00:11:30,870 Josh: So it seems like this is a really tough position to be bearish in or a tough 179 00:11:30,870 --> 00:11:31,910 Josh: time to bearish in at least. 180 00:11:32,330 --> 00:11:36,030 Ejaaz: Yeah, it's a it's incredibly risky trade to make. 181 00:11:36,310 --> 00:11:42,710 Ejaaz: And I mean, the cherry on top of all of this, Josh, is that he paid the price pretty dearly. 182 00:11:43,290 --> 00:11:47,570 Ejaaz: You sent a message to our group chat this morning and I had to read it like multiple times. 183 00:11:48,250 --> 00:11:54,030 Ejaaz: Michael Burry, this morning of our recording, announced that he is closing down 184 00:11:54,030 --> 00:11:59,470 Ejaaz: his fund, the very same fund which took out this $300 million short on NVIDIA. 185 00:11:59,630 --> 00:12:04,070 Ejaaz: He was down on his position pretty massively. He overstated his means. 186 00:12:04,230 --> 00:12:06,990 Ejaaz: And there's one line here, Josh, which kind of sums it all up. 187 00:12:07,270 --> 00:12:13,450 Ejaaz: He goes, my estimation of value in securities is not now and has not been for 188 00:12:13,450 --> 00:12:16,310 Ejaaz: some time in sync with the markets. 189 00:12:16,890 --> 00:12:20,410 Josh: Tough. Okay, so this connects the dots for me because I saw the beginning and I saw the end. 190 00:12:20,530 --> 00:12:24,070 Josh: I was like, okay, I know he goes out of business, but I do not know why or how 191 00:12:24,070 --> 00:12:26,230 Josh: he gets there. And I think this probably connects the story. 192 00:12:26,830 --> 00:12:29,330 Josh: Don't, again, to the point yesterday, don't bet against the optimist, 193 00:12:29,490 --> 00:12:31,990 Josh: dude. He had absolutely cooked. So where does that leave us then? 194 00:12:32,290 --> 00:12:37,510 Josh: So therefore, it is not a bubble and he was wrong or therefore it is a bubble 195 00:12:37,510 --> 00:12:38,770 Josh: and he was wrong with timing. 196 00:12:38,990 --> 00:12:44,390 Ejaaz: No, I like where you're sniffing josh because that something still seems a bit 197 00:12:44,390 --> 00:12:49,770 Ejaaz: off right it's like well okay if it's not gpus what could it potentially be also so. 198 00:12:49,770 --> 00:12:55,810 Josh: He he blew up his position but that doesn't ruin the hypothesis right like just 199 00:12:55,810 --> 00:13:00,190 Josh: the prices moved against him but it doesn't prove the thesis wrong is that right 200 00:13:00,190 --> 00:13:05,450 Ejaaz: It doesn't prove the thesis wrong on a long-term horizon because no one knows 201 00:13:05,450 --> 00:13:09,310 Ejaaz: whether any of these crazy spends are gonna be crazy in hindsight, 202 00:13:10,040 --> 00:13:14,320 Ejaaz: But for now, in the short term, all the fundamentals show that there is adequate 203 00:13:14,320 --> 00:13:18,320 Ejaaz: demand for the GPUs and there are end users that are willing to use these AI 204 00:13:18,320 --> 00:13:20,460 Ejaaz: products that require these GPUs. 205 00:13:20,660 --> 00:13:24,240 Ejaaz: It actually tells us the opposite story, which is there are not enough GPUs 206 00:13:24,240 --> 00:13:28,060 Ejaaz: in the world right now, old and new, that can satiate the demand for all the 207 00:13:28,060 --> 00:13:31,680 Ejaaz: AI products that are being served today, right now at this moment. 208 00:13:31,860 --> 00:13:32,400 Josh: Okay. Yeah. 209 00:13:33,780 --> 00:13:40,300 Ejaaz: But one thing that could signal a constraint, Josh, is energy. 210 00:13:40,900 --> 00:13:46,800 Ejaaz: The, put simply, the U.S. energy grid isn't up to par to supply energy to all 211 00:13:46,800 --> 00:13:48,640 Ejaaz: these GPUs so that they can do the job. 212 00:13:48,720 --> 00:13:52,780 Ejaaz: In fact, there's this clip that I have here from Satya Nadella of Microsoft 213 00:13:52,780 --> 00:13:57,160 Ejaaz: where he goes on to basically say he has hundreds of millions of dollars worth 214 00:13:57,160 --> 00:14:02,020 Ejaaz: of NVIDIA GPUs that are collecting dust in his fair weather data centers because 215 00:14:02,020 --> 00:14:04,420 Ejaaz: they don't have the energy to supply this. 216 00:14:04,420 --> 00:14:07,380 Josh: That seems uh troubling and i 217 00:14:07,380 --> 00:14:10,340 Josh: think this is to a point that is being made 218 00:14:10,340 --> 00:14:13,900 Josh: a lot where there are these gluts where 219 00:14:13,900 --> 00:14:18,300 Josh: they are and how they show themselves is going to be the thing for debate and 220 00:14:18,300 --> 00:14:22,460 Josh: this isn't the only hot clip we got this week there is this hysterical clip 221 00:14:22,460 --> 00:14:27,240 Josh: um of satya nadella ceo of microsoft kind of talking about his relationship 222 00:14:27,240 --> 00:14:30,140 Josh: with microsoft and i want to play this out before we give some commentary on 223 00:14:30,140 --> 00:14:32,860 Josh: it because i think of all the clips i watched this week. This was one of my favorite. 224 00:14:33,100 --> 00:14:38,420 Ejaaz: In our case, the good news here is OpenAI has a program which we have access to. 225 00:14:38,880 --> 00:14:43,300 Ejaaz: And so therefore, to think that Microsoft is not going to have something that's- What level of. 226 00:14:43,300 --> 00:14:44,120 Josh: Access do you have to that? 227 00:14:44,260 --> 00:14:44,660 Ejaaz: All of them. 228 00:14:44,940 --> 00:14:48,420 Josh: You just get the IP for all of that. So the only IP you don't have is a consumer 229 00:14:48,420 --> 00:14:50,280 Josh: hardware. That's it. Oh, wow. Okay. 230 00:14:53,860 --> 00:14:55,300 Ejaaz: Interesting. That's so good. That's crazy. 231 00:14:55,660 --> 00:14:58,760 Josh: And this comes off the back of Satya saying in a previous interview 232 00:14:58,760 --> 00:15:01,680 Josh: when asked about open ai he goes we are like above them we 233 00:15:01,680 --> 00:15:04,840 Josh: are beside them we are around them they have this full total 234 00:15:04,840 --> 00:15:07,760 Josh: controlling entity over open ai 235 00:15:07,760 --> 00:15:10,480 Josh: and this is it's like the delivery is 236 00:15:10,480 --> 00:15:13,880 Josh: hysterical the reaction is hysterical but it also it 237 00:15:13,880 --> 00:15:18,300 Josh: shows a lot of interesting dynamics between these large companies because during 238 00:15:18,300 --> 00:15:23,060 Josh: this interview satya also made mention of the fact that people who are building 239 00:15:23,060 --> 00:15:27,920 Josh: their moats around individual models are very fragile and very brittle in the 240 00:15:27,920 --> 00:15:33,860 Josh: sense that they are one company copying their model away from being worth significantly less. 241 00:15:34,040 --> 00:15:36,320 Josh: And we kind of saw that earlier this week with our Kimmy K2 episode. 242 00:15:36,440 --> 00:15:38,920 Josh: I highly recommend you watch it if you haven't, because 243 00:15:39,490 --> 00:15:45,950 Josh: It shows how fragile being a frontier model can be if that's your only business. 244 00:15:46,210 --> 00:15:50,190 Josh: And in the case of Satya and Microsoft, I mean, it seems like they have all 245 00:15:50,190 --> 00:15:51,010 Josh: the leverage in the world. 246 00:15:51,030 --> 00:15:54,010 Josh: If they want to fork the OpenAI code base right now and create their own chat 247 00:15:54,010 --> 00:15:56,590 Josh: GPT, they can do that. They own all of it. 248 00:15:56,710 --> 00:16:02,710 Ejaaz: Yeah, I've said this before, but I think the biggest winner of OpenAI is Satya 249 00:16:02,710 --> 00:16:04,710 Ejaaz: Nadella. He has struck the best deal. 250 00:16:06,330 --> 00:16:10,630 Ejaaz: OpenAI recently restructured their entire company. We actually did an episode on that. 251 00:16:10,970 --> 00:16:14,990 Ejaaz: Feel free to check it out. But one of the major takeaways from that was, 252 00:16:15,580 --> 00:16:22,240 Ejaaz: Satya owns 27% of OpenAI. And now he is able to engage in any model provider, not just OpenAI. 253 00:16:22,320 --> 00:16:26,200 Ejaaz: He doesn't have exclusive rights anymore. He can kind of like flirt with other 254 00:16:26,200 --> 00:16:27,520 Ejaaz: AI companies. And he did. 255 00:16:27,860 --> 00:16:30,440 Ejaaz: He engaged with like Amazon on a bunch of different things. 256 00:16:31,140 --> 00:16:36,300 Ejaaz: And he just owns all the IP until something like 2030 or 2032. 257 00:16:36,680 --> 00:16:39,500 Ejaaz: So he is one of the smartest chess players. 258 00:16:39,940 --> 00:16:43,440 Ejaaz: Honestly, unexpectedly for me, I didn't think I thought Microsoft was kind of 259 00:16:43,440 --> 00:16:46,660 Ejaaz: like boomer in this sets, but he's navigated it beautifully. 260 00:16:47,040 --> 00:16:51,440 Ejaaz: And one thing that I also need to give him kudos for is he has one of the biggest 261 00:16:51,440 --> 00:16:52,980 Ejaaz: moats when it comes to enterprise. 262 00:16:53,300 --> 00:16:56,960 Ejaaz: Like remember, Microsoft software is a really good consumer grade product, 263 00:16:57,180 --> 00:16:59,940 Ejaaz: but is mainly kind of making their money from all the enterprise stuff, 264 00:17:00,080 --> 00:17:01,560 Ejaaz: from Copala and stuff like that. 265 00:17:02,380 --> 00:17:04,860 Ejaaz: So beautifully executed from Satya here. 266 00:17:05,060 --> 00:17:07,820 Josh: I love this. Man, Satya, he's crushing it. 267 00:17:07,940 --> 00:17:12,040 Josh: And it's funny because you don't think of Microsoft as a serious player in the world of AI at all. 268 00:17:12,160 --> 00:17:15,080 Josh: I don't use any of their software. I don't use Copilot. In fact, 269 00:17:15,220 --> 00:17:19,360 Josh: he was asked about Copilot because Copilot was the single AI coding agent. 270 00:17:19,480 --> 00:17:20,980 Josh: It owned 100% of the market share. 271 00:17:21,140 --> 00:17:25,700 Josh: And now it's taken down to 25%. And he said, this is a great thing because now 272 00:17:25,700 --> 00:17:30,140 Josh: I still own 25% of a market that is now 10 times the size of what I previously 273 00:17:30,140 --> 00:17:33,280 Josh: owned 100% of. So for him, he's playing the positive sum game. 274 00:17:33,560 --> 00:17:38,100 Josh: He's very aware of the market dynamics at play. And for the cost of what was it, $10 billion? 275 00:17:38,620 --> 00:17:41,700 Josh: He got the largest shareholder position 276 00:17:41,700 --> 00:17:45,180 Josh: of open ai which was rumored to ipo at a trillion dollars 277 00:17:45,180 --> 00:17:49,800 Josh: so they not only see the financial upside but they get all the intellectual 278 00:17:49,800 --> 00:17:53,100 Josh: property most importantly the code base to either reverse engineer for their 279 00:17:53,100 --> 00:17:58,020 Josh: own wants or to just clone and use for whatever they want it for it they have 280 00:17:58,020 --> 00:18:03,840 Josh: it and it is a true chess move by satya bravo nicely done microsoft as 281 00:18:03,840 --> 00:18:09,940 Ejaaz: Of today as we speak today at open ai's 500 billion dollar valuation uh microsoft, 282 00:18:10,770 --> 00:18:14,530 Ejaaz: Microsoft stake is worth $136 billion. 283 00:18:15,390 --> 00:18:22,190 Josh: Not a bad deal. But that's not the only big news. We have exciting news on the ChatGPT front. 284 00:18:22,410 --> 00:18:26,430 Josh: This is what I'm most excited about because I use ChatGPT every day and this 285 00:18:26,430 --> 00:18:30,910 Josh: is hopefully going to change the way I use it. Maybe we'll see when we talk about it. 286 00:18:31,170 --> 00:18:36,130 Josh: The news this week is that ChatGPT and OpenAI, they launched GPT 5.1, 287 00:18:36,450 --> 00:18:38,830 Josh: which is a whole new upgrade to GPT-5. 288 00:18:39,530 --> 00:18:42,630 Josh: How different it is, we're not really sure. 289 00:18:42,850 --> 00:18:46,690 Josh: I know, EJ, as I kind of read through the highlights, it's warmer in terms of 290 00:18:46,690 --> 00:18:50,090 Josh: its like sentiment by default. So it'll be a little bit nicer, 291 00:18:50,350 --> 00:18:54,270 Josh: which is bizarre because I thought we were going, we were trying to go the opposite 292 00:18:54,270 --> 00:18:55,870 Josh: direction, like it's a little too nice. 293 00:18:56,110 --> 00:18:58,430 Josh: And they added some safeguards that we'll get into in a second. 294 00:18:58,610 --> 00:19:02,530 Josh: There's also much better instruction following, which is interesting because 295 00:19:02,530 --> 00:19:07,450 Josh: oftentimes when you instruct GPT-5 to do things, it doesn't always do them exactly as you want. 296 00:19:07,650 --> 00:19:12,170 Josh: So like if you say always respond in six words, it will oftentimes respond in 297 00:19:12,170 --> 00:19:14,990 Josh: more or less than six words. It just doesn't quite understand the instructions. 298 00:19:15,270 --> 00:19:20,650 Josh: And then one of the really fun things that I saw was they changed the different 299 00:19:20,650 --> 00:19:21,630 Josh: types of personalities. 300 00:19:21,790 --> 00:19:25,350 Josh: So you can choose between personalities on how you engage with the model. 301 00:19:25,470 --> 00:19:29,350 Josh: And the previous ones were default, friendly, and efficient. 302 00:19:29,920 --> 00:19:32,940 Josh: And now there are some newly added ones, which are professional, 303 00:19:33,260 --> 00:19:35,760 Josh: candid, quirky, nerdy, and cynical. 304 00:19:36,020 --> 00:19:41,340 Josh: So in the case you want to personalize ChatGPT to be more like those models, 305 00:19:41,720 --> 00:19:45,660 Josh: well, now you have the option to do so with GPT 5.1. 306 00:19:45,780 --> 00:19:49,840 Josh: So if I'm sick of it saying, you're so right, I agree, that's a great idea, 307 00:19:50,000 --> 00:19:52,580 Josh: that's a great question, you could just say, cut that out, and it'll actually 308 00:19:52,580 --> 00:19:54,620 Josh: listen for the first time, which is really exciting. 309 00:19:55,020 --> 00:19:58,460 Josh: EJS, did you find anything interesting from GPT 5.1 after going through everything? 310 00:19:58,460 --> 00:20:01,960 Ejaaz: Yeah, I spent the last like 12 hours playing around with it. 311 00:20:02,200 --> 00:20:06,500 Ejaaz: Can I just, I'm going to play Bad Cop for a bit, Josh. Cool. I don't care. 312 00:20:06,820 --> 00:20:10,220 Ejaaz: Like this is kind of like a nothing burger 0.1 update. 313 00:20:11,120 --> 00:20:14,200 Ejaaz: And it's honestly like I've come to expect more from open air. 314 00:20:14,260 --> 00:20:16,920 Ejaaz: So I'm kind of surprised that they've come through with this release. 315 00:20:17,260 --> 00:20:20,960 Ejaaz: On the point of different personalities, I think that's super helpful. 316 00:20:21,220 --> 00:20:27,580 Ejaaz: Because speaking for myself, I think that the current model is too agreeable. It's too sycophantic. 317 00:20:28,660 --> 00:20:34,120 Ejaaz: And kind of Sam heard this feedback and the original version of GPT-5 was less 318 00:20:34,120 --> 00:20:36,940 Ejaaz: sycophantic, but then a bunch of people were like, I want it to be more friendly. 319 00:20:37,060 --> 00:20:40,480 Ejaaz: I want it to be more appeasing. And so he's kind of like flittering back and forth. 320 00:20:41,180 --> 00:20:44,720 Ejaaz: This option gives a lot more flexibility for the end user, for me. 321 00:20:44,840 --> 00:20:48,500 Ejaaz: I'm like, I want someone that's maybe a bit more candid or a little more cynical 322 00:20:48,500 --> 00:20:51,200 Ejaaz: when I'm kind of doing my research and then maybe a little more friendly when 323 00:20:51,200 --> 00:20:52,660 Ejaaz: I talk about more personal stuff, right? 324 00:20:52,920 --> 00:20:55,480 Ejaaz: So the pliability is cool. 325 00:20:55,980 --> 00:20:58,720 Ejaaz: What I will say is like, you kind of were able to do 326 00:20:58,720 --> 00:21:01,920 Ejaaz: this before you could just go into settings user personalization 327 00:21:01,920 --> 00:21:04,780 Ejaaz: and type in a prompt this i guess makes it easier 328 00:21:04,780 --> 00:21:07,640 Ejaaz: because you just gotta click a button and maybe that helps a lot of people 329 00:21:07,640 --> 00:21:12,380 Ejaaz: um the other cool thing i guess is like uh when it comes to efficiency it uses 330 00:21:12,380 --> 00:21:17,240 Ejaaz: fewer tokens and gives you more output so you get a higher quality answer for 331 00:21:17,240 --> 00:21:20,640 Ejaaz: less energy for less compute which is probably going to drive down costs of 332 00:21:20,640 --> 00:21:24,920 Ejaaz: accessing this type of aid this type of Frontier AI, which is super cool. 333 00:21:25,180 --> 00:21:28,320 Ejaaz: But aside from that, it's kind of like, why did they do this? 334 00:21:28,480 --> 00:21:31,000 Ejaaz: One of the critiques I saw, Josh is like, 335 00:21:31,440 --> 00:21:35,860 Ejaaz: it seems kind of rushed. Like there's no benchmarks, no API release. 336 00:21:36,720 --> 00:21:40,480 Ejaaz: A bunch of the dev tools look a little sloppy. It's almost like they kind of rush this. 337 00:21:40,620 --> 00:21:44,620 Ejaaz: It's not, there's no formal kind of like Sam's doing a live stream around this. 338 00:21:44,760 --> 00:21:47,660 Ejaaz: It's nothing around like, you know, hey, here's how it compares against other 339 00:21:47,660 --> 00:21:50,820 Ejaaz: models. They just kind of like rush this out. And I'm kind of confused. 340 00:21:51,080 --> 00:21:55,120 Josh: Yeah. I suspect the public sentiment will start to reshape itself around how this goes. 341 00:21:55,600 --> 00:21:58,300 Josh: Like when, when Apple releases iOS 26, it 342 00:21:58,300 --> 00:22:01,460 Josh: was a huge release with liquid glass and all this cool new improvements and 343 00:22:01,460 --> 00:22:04,180 Josh: then 26.1 is an incremental improvement where it adds a lot 344 00:22:04,180 --> 00:22:07,300 Josh: of new features nothing really groundbreaking but the product gets light slightly 345 00:22:07,300 --> 00:22:10,420 Josh: better i think we can kind of view these mid-tier updates 346 00:22:10,420 --> 00:22:13,120 Josh: as that where they're just small incremental improvements like 347 00:22:13,120 --> 00:22:15,980 Josh: now it's a little more efficient like you said there's this 348 00:22:15,980 --> 00:22:18,680 Josh: new dynamic thinking time which is what they call like 349 00:22:18,680 --> 00:22:21,940 Josh: smart thinking and it allows it to think faster 350 00:22:21,940 --> 00:22:25,160 Josh: think more efficiently use less words that 351 00:22:25,160 --> 00:22:27,880 Josh: are fillers so it needs less tokens to operate like the chart that we're 352 00:22:27,880 --> 00:22:30,900 Josh: seeing and that just allows for one the 353 00:22:30,900 --> 00:22:33,700 Josh: model to think more because it uses less tokens to think 354 00:22:33,700 --> 00:22:37,120 Josh: and then two it just creates a little more efficiency unlock for open 355 00:22:37,120 --> 00:22:41,200 Josh: ai's gpus to open them up to do other things so i assume this update is probably 356 00:22:41,200 --> 00:22:45,700 Josh: for both people for the consumer and for open ai in terms of efficiency and 357 00:22:45,700 --> 00:22:49,240 Josh: slightly better product is it this amazing novel breakthrough it doesn't appear 358 00:22:49,240 --> 00:22:53,920 Josh: like it at all in fact it's just like a very marginal improvement but it's something it is like 359 00:22:54,460 --> 00:22:57,680 Josh: It is something, and I suspect we'll probably get more updates like this, 360 00:22:57,780 --> 00:23:00,000 Josh: where it's like, oh, well, it does this like one thing a little bit better. 361 00:23:00,360 --> 00:23:04,060 Josh: Do you notice it? Maybe like one in every 50 prompts, but it's not a big deal. 362 00:23:04,580 --> 00:23:07,440 Josh: And that's probably where we are for a little while until maybe Gemini 3. 363 00:23:07,680 --> 00:23:11,920 Josh: I don't know. We have, there are murmurs of Gemini 3 coming down the pipeline from Google. 364 00:23:11,920 --> 00:23:13,720 Ejaaz: I've been waiting for it for a month now. 365 00:23:14,020 --> 00:23:16,100 Josh: Yeah, that could hopefully blow things out of the water. But we're at this weird 366 00:23:16,100 --> 00:23:19,700 Josh: kind of stagnant period where we haven't seen those step function improvements 367 00:23:19,700 --> 00:23:21,700 Josh: in AI models in a little while now. 368 00:23:21,700 --> 00:23:25,640 Ejaaz: I mean, like the speculation that would be like we're kind of grinding to a 369 00:23:25,640 --> 00:23:30,400 Ejaaz: halt and that exponential up, that S curve where we see AI improving massively 370 00:23:30,400 --> 00:23:34,040 Ejaaz: and it's curing cancer may not be quite there yet. 371 00:23:34,160 --> 00:23:38,500 Ejaaz: But hey, listen, in the meantime, there are several parties that are super bullish 372 00:23:38,500 --> 00:23:42,600 Ejaaz: open AI. One of them is Masayoshi-san of SoftBank. 373 00:23:43,380 --> 00:23:51,900 Ejaaz: In this news, he sold his entire stake of NVIDIA worth $5.83 billion to buy OpenAI stock. 374 00:23:52,040 --> 00:23:55,960 Ejaaz: It's hilarious because all that money that is investing in OpenAI is only going 375 00:23:55,960 --> 00:23:57,760 Ejaaz: to get spent on NVIDIA GPUs. 376 00:23:59,340 --> 00:24:02,520 Ejaaz: Just hold NVIDIA. Like, why are you doing that? 377 00:24:02,840 --> 00:24:05,540 Josh: Masa's really, he's been making some questionable decisions lately. 378 00:24:05,540 --> 00:24:08,800 Josh: He sold NVIDIA too early and he missed the whole bubble he missed the whole 379 00:24:08,800 --> 00:24:12,240 Josh: like he's just been making some strange decisions here he 380 00:24:12,240 --> 00:24:18,740 Ejaaz: Wants to own 5% of NVIDIA do you know how much that would be worth today? I cannot believe, 381 00:24:19,590 --> 00:24:21,890 Ejaaz: So like, you know, someone like kind of like tweeted this. They were like, 382 00:24:21,970 --> 00:24:25,690 Ejaaz: you know, someone should look into what happened after he sold his entire NVIDIA stake in 2019. 383 00:24:26,070 --> 00:24:32,310 Ejaaz: Just for reference here, the GPT 3.5 that went viral across the internet released 384 00:24:32,310 --> 00:24:33,570 Ejaaz: two and a half years later. 385 00:24:33,970 --> 00:24:37,710 Ejaaz: So pretty, pretty insane thing there. And then in final news, 386 00:24:38,110 --> 00:24:40,590 Ejaaz: Josh, tell me about this. This looks super exciting. 387 00:24:40,850 --> 00:24:44,250 Josh: The Google DeepMind team is back with some pretty awesome news. 388 00:24:44,430 --> 00:24:47,590 Josh: We are very fond of the Google DeepMind team in particular here at Limitless. 389 00:24:47,590 --> 00:24:52,930 Josh: They are very good at building real world physics in a digital manifestation, 390 00:24:52,930 --> 00:24:56,850 Josh: which I think is a really important thing to be good at in a world where we're 391 00:24:56,850 --> 00:25:01,450 Josh: training robots to be good at physical stuff, but they have not quite had the 392 00:25:01,450 --> 00:25:03,370 Josh: training time to get good at physical stuff. 393 00:25:03,490 --> 00:25:06,850 Josh: And what I love is that we've seen a lot of these projects coming out of the 394 00:25:06,850 --> 00:25:08,390 Josh: DeepMind team. And today we got a 395 00:25:10,130 --> 00:25:13,630 Josh: An agent that is capable of navigating these 3D virtual worlds. 396 00:25:13,750 --> 00:25:16,250 Josh: So we saw previously there was a few models that can generate the worlds. 397 00:25:16,370 --> 00:25:19,690 Josh: Well, now SEMA 2, which implies there was a SEMA 1 that I was blissfully unaware 398 00:25:19,690 --> 00:25:24,570 Josh: of, allows these AI agents to actually navigate these virtual worlds. 399 00:25:24,690 --> 00:25:28,490 Josh: So now not only can Google generate the virtual games, they literally look like 400 00:25:28,490 --> 00:25:32,470 Josh: video games in this in this video, but they can also have the characters navigate 401 00:25:32,470 --> 00:25:37,290 Josh: these complicated experiences, understand how things work, kind of piece things together. 402 00:25:37,550 --> 00:25:40,310 Josh: It's amazing to see because it feels like as i'm watching 403 00:25:40,310 --> 00:25:43,330 Josh: this video it is how a toddler would play 404 00:25:43,330 --> 00:25:46,450 Josh: a video game it's kind of they're slowly moving it's figuring 405 00:25:46,450 --> 00:25:49,510 Josh: it out you could see kind of reasoning in real time and navigating 406 00:25:49,510 --> 00:25:52,930 Josh: these spaces and it's important to understand as you're watching this everything 407 00:25:52,930 --> 00:25:56,590 Josh: that is on this is generated by ai because previously 408 00:25:56,590 --> 00:25:59,690 Josh: these look like trailers to video games and those those video 409 00:25:59,690 --> 00:26:03,210 Josh: games would have required thousands of people who are game developers lots 410 00:26:03,210 --> 00:26:05,890 Josh: of time on the game engine and all of that is 411 00:26:05,890 --> 00:26:08,750 Josh: done now fully by ai so seeing this announcement 412 00:26:08,750 --> 00:26:11,690 Josh: was really exciting for me because i'm like one i just love video games and 413 00:26:11,690 --> 00:26:15,710 Josh: two like oh my god wait these ais not only are they able to generate the environment 414 00:26:15,710 --> 00:26:19,390 Josh: but now these ai models are able to actually engage with the environment and 415 00:26:19,390 --> 00:26:23,590 Josh: what downstream effects does that have on training humanoid robots and the like 416 00:26:23,590 --> 00:26:26,370 Josh: which we saw with the tesla episode last week yeah 417 00:26:26,370 --> 00:26:31,770 Ejaaz: I mean yeah to your point um we see a real life implementation of this with 418 00:26:31,770 --> 00:26:36,870 Ejaaz: tesla tesla Tesla has a world model which they use in their robots, 419 00:26:37,050 --> 00:26:42,110 Ejaaz: which is the Tesla cars and automated self-driving and the future optimist robots. 420 00:26:42,250 --> 00:26:46,610 Ejaaz: And what's cool about this is, Josh, one of the main things that makes your 421 00:26:46,610 --> 00:26:51,230 Ejaaz: AI models super cool or really smart or intelligent is data. 422 00:26:51,980 --> 00:26:55,980 Ejaaz: There's a scarcity of data. There's a scarcity of rich human data. 423 00:26:56,140 --> 00:26:59,620 Ejaaz: Like they don't necessarily see what we see through our eyes or hear what we hear. 424 00:26:59,800 --> 00:27:04,940 Ejaaz: These simulated environments basically synthetically create this data and enhance 425 00:27:04,940 --> 00:27:06,460 Ejaaz: any of these AI models, right? 426 00:27:06,760 --> 00:27:11,320 Ejaaz: And so Tesla kind of gets smarter by replaying a scenario where it almost crashed 427 00:27:11,320 --> 00:27:15,720 Ejaaz: over and over and over again in simulated environments until it gets it right. 428 00:27:15,880 --> 00:27:20,940 Ejaaz: And then that just gets transported to every single production ready Tesla that 429 00:27:20,940 --> 00:27:24,780 Ejaaz: anyone is driving so that if they found themselves in a similar scenario, they avoid it. 430 00:27:24,860 --> 00:27:28,760 Ejaaz: And that's just like one small use case for like what these robots will eventually 431 00:27:28,760 --> 00:27:32,400 Ejaaz: be benefiting from. So such a super cool update. 432 00:27:32,580 --> 00:27:35,300 Josh: Yeah. And if you scroll down actually to post EJs, you'll see a lot of similarities 433 00:27:35,300 --> 00:27:39,800 Josh: to what we spoke about with Tesla, where you could see the model kind of reasoning in plain English. 434 00:27:39,980 --> 00:27:42,680 Josh: You could see, oh, we equipped the pickaxe and copper mine. And then, 435 00:27:42,800 --> 00:27:45,560 Josh: okay, like I'll equip the pickaxe and it walks you through its chain of thought 436 00:27:45,560 --> 00:27:48,580 Josh: and then not only that but the post below it shows that it's self-improving 437 00:27:48,580 --> 00:27:52,360 Josh: so the more time it spends cycling through the more time it spends playing games 438 00:27:52,360 --> 00:27:56,280 Josh: the better it gets and these are early implementations of what what we kind 439 00:27:56,280 --> 00:27:59,760 Josh: of imagined ai would eventually be which is this self-recursive loop where it 440 00:27:59,760 --> 00:28:04,000 Josh: can improve without outside information and it's really exciting to see it happening 441 00:28:04,620 --> 00:28:08,120 Josh: in the virtual world, because I guess that's the only place it can make sense 442 00:28:08,120 --> 00:28:09,480 Josh: where it costs much less. 443 00:28:09,680 --> 00:28:13,400 Josh: You can't harm anybody if a robot goes rogue, but it can learn without the inputs 444 00:28:13,400 --> 00:28:17,220 Josh: of other people. So that is the exciting thing about this, I think. 445 00:28:17,580 --> 00:28:20,300 Josh: And I'm really excited for the DeepMind team, man. I hope they keep going. 446 00:28:20,760 --> 00:28:24,480 Ejaaz: Well, dude, like, do you remember our conversation with Logan Kilpatrick, 447 00:28:24,680 --> 00:28:26,560 Ejaaz: the head of Google's AI studio? 448 00:28:26,840 --> 00:28:28,860 Ejaaz: If you haven't seen that episode, definitely go and check it out. 449 00:28:29,120 --> 00:28:34,040 Ejaaz: One of his main takeaways or one of my main takeaways is all of Google's AI 450 00:28:34,040 --> 00:28:38,220 Ejaaz: byproducts feeds off of different models within the same suite. 451 00:28:38,400 --> 00:28:43,160 Ejaaz: So if their video model learns something cool, they can transpose that onto 452 00:28:43,160 --> 00:28:47,820 Ejaaz: their LLM, which is like in Word, so that it gets recursively smarter over time. 453 00:28:47,960 --> 00:28:52,960 Ejaaz: I just think it's like fascinating and world models kind of like manifest all 454 00:28:52,960 --> 00:28:56,260 Ejaaz: of these different things in one particular simulated environment. 455 00:28:56,680 --> 00:29:00,020 Ejaaz: So, so cool. Kind of hard to wrap my head around sometimes, but because it looks 456 00:29:00,020 --> 00:29:01,740 Ejaaz: like a video game to your point but but awesome. 457 00:29:01,740 --> 00:29:06,140 Josh: Yeah it's you have to break this gap between like your normal understanding 458 00:29:06,140 --> 00:29:09,000 Josh: when you see these videos and think like oh my god wait this is not normal this 459 00:29:09,000 --> 00:29:12,940 Josh: is actually all ai generated um but with that we have concluded all the fun 460 00:29:12,940 --> 00:29:14,300 Josh: new updates from this week it was a 461 00:29:14,740 --> 00:29:18,140 Josh: i'd say a medium week it wasn't anything crazy there was no crazy outliers but 462 00:29:18,140 --> 00:29:21,660 Josh: a solid week overall every week there's so much stuff going on and the progress 463 00:29:21,660 --> 00:29:25,500 Josh: i feel like we're almost getting immune to it how fast things are moving um 464 00:29:25,500 --> 00:29:29,800 Josh: but i hope everyone found this interesting if you did please don't remember 465 00:29:29,800 --> 00:29:31,000 Josh: to share with your friends, like, subscribe. 466 00:29:31,160 --> 00:29:35,080 Josh: Ejaz, I know you were manifesting some sort of send-off today. Do you want to share? 467 00:29:35,280 --> 00:29:41,580 Ejaaz: Yes. Similar to Michael Burry, you shouldn't be shorting the biggest podcast bubble in the world. 468 00:29:41,680 --> 00:29:46,480 Ejaaz: And Limitless is at the forefront over here. We're delivering you all the latest AI news. 469 00:29:46,700 --> 00:29:51,420 Ejaaz: And just like the trade that Burry made that he lost $300 million in, 470 00:29:51,760 --> 00:29:54,900 Ejaaz: you don't want to be shorting the best leading AI podcast in the world. 471 00:29:55,100 --> 00:29:56,340 Ejaaz: Give us a five-star rating. 472 00:29:56,680 --> 00:30:00,140 Ejaaz: Subscribe if you're not. What was the stat, Josh? 80% of listeners? 473 00:30:00,400 --> 00:30:04,280 Ejaaz: 83%? How many of them don't subscribe? Over 80%. 474 00:30:04,650 --> 00:30:07,950 Ejaaz: Over 80. That's nuts. It helps us out massively. 475 00:30:08,250 --> 00:30:12,390 Ejaaz: We are currently in the top 30 in the technology podcast across Spotify, 476 00:30:12,550 --> 00:30:15,990 Ejaaz: Apple, and wherever you listen. Help us get to the top 10. It would mean a lot. What do we say? 477 00:30:16,470 --> 00:30:20,230 Josh: Don't ever bet against the optimists. And by subscribing and by being a part 478 00:30:20,230 --> 00:30:21,730 Josh: of this, you are part of the optimists. 479 00:30:21,870 --> 00:30:24,970 Josh: And we are going to move forward and things are going to progress. 480 00:30:25,210 --> 00:30:29,150 Josh: And just don't bet against the optimists. So thank you for being optimistic 481 00:30:29,150 --> 00:30:33,050 Josh: with us and being on this journey with us and we'll be back next week with a 482 00:30:33,050 --> 00:30:36,450 Josh: whole new slew of episodes so thank you for watching and we'll see you then