WEBVTT 00:00:00.000 --> 00:00:02.080 Imagine the biggest construction project of our 00:00:02.080 --> 00:00:05.299 lifetime. It's not about sprawling rail lines, 00:00:05.500 --> 00:00:08.140 you know, crisscrossing continents. Right. Nor 00:00:08.140 --> 00:00:11.220 is it about like an intricate web of new highways. 00:00:11.460 --> 00:00:14.500 This one is largely unseen, yet its scale, its 00:00:14.500 --> 00:00:17.600 sheer ambition. Well. It's absolutely staggering. 00:00:17.739 --> 00:00:19.660 Yeah, we're talking about AI compute, aren't 00:00:19.660 --> 00:00:21.460 we? It's essentially the digital backbone of 00:00:21.460 --> 00:00:25.379 the future, the very infrastructure powering 00:00:25.379 --> 00:00:28.300 this new technological era. Exactly. But here's 00:00:28.300 --> 00:00:30.940 the really wild part, I think. The question that's 00:00:30.940 --> 00:00:34.280 just sort of begging to be asked, who is actually 00:00:34.280 --> 00:00:37.630 footing this well? Mind -boggling, Bill. Welcome 00:00:37.630 --> 00:00:40.109 to the Deep Dive. This is where we take the information 00:00:40.109 --> 00:00:42.850 you've shared with us and unpack it, seeking 00:00:42.850 --> 00:00:45.450 those crucial nuggets of knowledge. Today, we're 00:00:45.450 --> 00:00:48.250 diving deep into the fascinating intersection 00:00:48.250 --> 00:00:52.250 of AI's booming infrastructure and, well, a noticeable 00:00:52.250 --> 00:00:54.609 shift in the industry's focus. You've given us 00:00:54.609 --> 00:00:56.770 some truly insightful source material, a fantastic 00:00:56.770 --> 00:00:59.189 newsletter to explore. That's right. We're going 00:00:59.189 --> 00:01:01.450 to journey through the monumental financial investment 00:01:01.450 --> 00:01:05.129 behind AI, looking at the just incredible sums 00:01:05.129 --> 00:01:08.269 pouring into data centers. Then we'll pivot a 00:01:08.269 --> 00:01:10.709 bit to some of the surprisingly clever and practical 00:01:10.709 --> 00:01:13.530 new AI capabilities that are emerging right now. 00:01:13.629 --> 00:01:16.469 Okay. And finally, we'll unpack why so many are 00:01:16.469 --> 00:01:19.030 saying we're entering what's being called the 00:01:19.030 --> 00:01:21.730 post -hype era for AI. Kind of interesting. A 00:01:21.730 --> 00:01:24.030 mission, as always, is to extract the most important 00:01:24.030 --> 00:01:26.549 insights, give you that valuable shortcut to 00:01:26.549 --> 00:01:29.530 being truly well -informed, and hopefully spark 00:01:29.530 --> 00:01:33.329 a few aha moments along the way. Yeah. So let's 00:01:33.329 --> 00:01:35.209 unpack this. Okay, we really have to start with 00:01:35.209 --> 00:01:39.159 what feels like the $3 trillion question. Who's 00:01:39.159 --> 00:01:41.459 actually paying for this unprecedented AI building 00:01:41.459 --> 00:01:44.060 boom? Right. Because when you look at the projections, 00:01:44.280 --> 00:01:46.299 it becomes clear this is probably the single 00:01:46.299 --> 00:01:49.120 biggest capital project of our lifetime. A truly 00:01:49.120 --> 00:01:51.140 generational undertaking. It really is. And the 00:01:51.140 --> 00:01:54.400 numbers are just, wow, jaw dropping. Morgan Stanley, 00:01:54.579 --> 00:01:56.319 for instance, projects were hurtling towards 00:01:56.319 --> 00:01:59.599 an incredible $2 .9 trillion in AI data center 00:01:59.599 --> 00:02:02.739 spend by 2029. $2 .9 trillion. Yeah. And to put 00:02:02.739 --> 00:02:05.760 that into perspective, big tech giants like Google, 00:02:05.920 --> 00:02:08.930 Amazon, Microsoft and Meta alone. They're set 00:02:08.930 --> 00:02:11.849 to spend a combined $400 billion on data centers 00:02:11.849 --> 00:02:14.810 in a single year. Just two years from now, by 00:02:14.810 --> 00:02:18.289 2026, that is an immense number of servers, an 00:02:18.289 --> 00:02:21.009 absolute forest of them, you know, humming away. 00:02:21.310 --> 00:02:22.750 And here's where it gets really interesting, 00:02:22.849 --> 00:02:25.330 maybe a little bit concerning, too. Our sources 00:02:25.330 --> 00:02:28.439 indicate that big tech companies. They're projected 00:02:28.439 --> 00:02:31.740 to cover only about half of that $2 .9 trillion 00:02:31.740 --> 00:02:34.319 bill. Only half. Yeah. That leaves a massive, 00:02:34.379 --> 00:02:37.500 almost incomprehensible $1 .5 trillion funding 00:02:37.500 --> 00:02:39.819 hole. Good grief. And these aren't your average 00:02:39.819 --> 00:02:41.520 server farms, right? We're talking about projects 00:02:41.520 --> 00:02:44.659 with names like Metis Prometheus, XAI's Colossus, 00:02:44.659 --> 00:02:48.199 or OpenAI's Stargate, each one reportedly costing 00:02:48.199 --> 00:02:51.259 over $100 billion. Each one. These aren't small 00:02:51.259 --> 00:02:53.560 -scale ventures. They're like megastructures 00:02:53.560 --> 00:02:55.689 of digital processing power. What's absolutely 00:02:55.689 --> 00:02:57.449 fascinating here is how they're filling that 00:02:57.449 --> 00:03:00.110 hole. The short answer, straight from the source 00:03:00.110 --> 00:03:04.849 material, is borrow like crazy. In 2025 alone, 00:03:05.050 --> 00:03:08.229 a staggering $60 billion in loans are already 00:03:08.229 --> 00:03:10.930 funneled into roughly $440 billion worth of these 00:03:10.930 --> 00:03:14.550 massive projects. And to give a concrete example, 00:03:14.710 --> 00:03:17.050 Meta's enormous data centers in Ohio and Louisiana 00:03:17.050 --> 00:03:20.849 collectively raised $29 billion. $29 billion. 00:03:21.090 --> 00:03:24.469 And an astonishing $26 billion of that, the vast 00:03:24.469 --> 00:03:28.449 majority, was pure debt. This isn't just tech 00:03:28.449 --> 00:03:31.550 companies spending their deep pockets. This is 00:03:31.550 --> 00:03:34.750 Wall Street's cash leveraged to the hilt, pouring 00:03:34.750 --> 00:03:37.849 into what is essentially a brand new type of 00:03:37.849 --> 00:03:41.110 global infrastructure. But this certainly raises 00:03:41.110 --> 00:03:43.370 some very important questions, doesn't it? Absolutely. 00:03:43.629 --> 00:03:45.810 Like what are the red flags nobody seems to be 00:03:45.810 --> 00:03:48.110 talking about in polite company? What are the 00:03:48.110 --> 00:03:51.409 risks inherent in this model? Well, this whole. 00:03:52.120 --> 00:03:54.479 Situation brings up a critical question. What 00:03:54.479 --> 00:03:57.379 are the true risks lurking beneath this investment 00:03:57.379 --> 00:03:59.840 frenzy? Our sources highlight several key ones. 00:03:59.960 --> 00:04:03.060 OK. We see the risk of overcapacity, right, where 00:04:03.060 --> 00:04:06.900 the buildout exceeds actual demand. There's significant 00:04:06.900 --> 00:04:10.460 obsolescence risk given how quickly AI tech evolves. 00:04:10.819 --> 00:04:13.099 That stuff could be outdated fast. Yeah, really 00:04:13.099 --> 00:04:15.319 fast. And then there's exit risk, meaning, you 00:04:15.319 --> 00:04:17.899 know, the difficulty of unwinding these huge 00:04:17.899 --> 00:04:20.000 specialized investments if they don't pan out. 00:04:20.360 --> 00:04:22.480 When we connect this to the bigger picture, it 00:04:22.480 --> 00:04:25.759 feels like deja vu. We've seen this before with 00:04:25.759 --> 00:04:27.920 past infrastructure bubbles like the frenzied 00:04:27.920 --> 00:04:30.100 railroad expansion of the 19th century. Oh, yeah. 00:04:30.199 --> 00:04:33.959 The dot -com boom and bust or the telecom overbuild 00:04:33.959 --> 00:04:36.740 of the early 2000s. It's kind of the same old 00:04:36.740 --> 00:04:40.379 story just with new GPUs and, well, vastly higher 00:04:40.379 --> 00:04:43.339 stakes. So instead of slowing down or, you know, 00:04:43.379 --> 00:04:46.079 self -funding entirely, these companies are essentially 00:04:46.079 --> 00:04:48.720 outsourcing the risk. Mm -hmm. That's a good 00:04:48.720 --> 00:04:50.439 way to put it. They're letting private capital 00:04:50.439 --> 00:04:53.680 and ultimately debt markets absorb the downside 00:04:53.680 --> 00:04:56.879 if demand for AI compute doesn't quite match 00:04:56.879 --> 00:04:59.959 the, well, gargantuan build out we're seeing. 00:05:00.079 --> 00:05:02.060 It's an interesting maneuver. So what does this 00:05:02.060 --> 00:05:04.199 all mean for the broader financial system? It's 00:05:04.199 --> 00:05:06.079 arguably one of the biggest financial system 00:05:06.079 --> 00:05:09.050 stress tests in decades. Wow. It implies that 00:05:09.050 --> 00:05:11.470 AI's biggest hurdles, its biggest risks might 00:05:11.470 --> 00:05:13.550 not be technical breakthroughs or safety concerns 00:05:13.550 --> 00:05:17.050 at all, but rather they could be purely financial 00:05:17.050 --> 00:05:21.689 and systemic. Pete, whoa. Imagine scaling to 00:05:21.689 --> 00:05:24.029 a billion queries or maybe, thinking about it 00:05:24.029 --> 00:05:26.329 now, a billion dollars just to build the basic 00:05:26.329 --> 00:05:29.370 infrastructure. It's truly mind -boggling how 00:05:29.370 --> 00:05:32.839 much... Capital is tied up in something so foundational 00:05:32.839 --> 00:05:36.040 yet, you know, so unproven in terms of long -term 00:05:36.040 --> 00:05:38.860 return. That's a powerful statement. Yes. It 00:05:38.860 --> 00:05:42.100 truly makes you realize the sheer scale of this 00:05:42.100 --> 00:05:44.980 undertaking and how the risks are involving beyond 00:05:44.980 --> 00:05:48.399 just the tech itself. Okay. So if we're stepping 00:05:48.399 --> 00:05:50.639 back from these incredible numbers and this new 00:05:50.639 --> 00:05:53.939 financial model, what's the single biggest takeaway 00:05:53.939 --> 00:05:56.079 for our listeners from this massive investment 00:05:56.079 --> 00:05:58.550 strategy? I'd say the biggest takeaway is that 00:05:58.550 --> 00:06:01.470 this is a colossal financial gamble, heavily 00:06:01.470 --> 00:06:04.389 leveraged by Wall Street's cash, making the entire 00:06:04.389 --> 00:06:07.250 financial system deeply intertwined with AI's 00:06:07.250 --> 00:06:10.269 future. Okay. A colossal gamble. Got it. Let's 00:06:10.269 --> 00:06:11.970 pivot now from the colossal financial picture 00:06:11.970 --> 00:06:15.029 to some of the really cool specific developments 00:06:15.029 --> 00:06:17.129 and capabilities emerging in AI. Yeah, let's 00:06:17.129 --> 00:06:18.750 do it. Because while the infrastructure story 00:06:18.750 --> 00:06:22.089 is immense, the practical progress is also incredibly 00:06:22.089 --> 00:06:24.329 rapid. Absolutely. On the creative front, for 00:06:24.329 --> 00:06:26.930 example, Google's Megan for image. generators 00:06:26.930 --> 00:06:28.970 now generally available, which is pretty neat 00:06:28.970 --> 00:06:33.189 for creators. Yeah, it offers up to 2K resolution, 00:06:33.550 --> 00:06:36.350 so really high quality, and it has a faster model 00:06:36.350 --> 00:06:38.410 for quicker outputs. You can generate stunning 00:06:38.410 --> 00:06:40.709 visuals almost instantly. And for the gamers 00:06:40.709 --> 00:06:44.230 among you, there was this viral test on X that 00:06:44.230 --> 00:06:47.769 showed GPT -5, the latest from OpenAI, earning 00:06:47.769 --> 00:06:51.870 8 badges in the classic Pokemon Red video game 00:06:51.870 --> 00:06:55.079 in just 6 ,000 steps. Now, That might sound like 00:06:55.079 --> 00:06:58.139 a lot of steps. Yeah. But it's almost 70 % fewer 00:06:58.139 --> 00:07:00.100 steps than a model that came out only six months 00:07:00.100 --> 00:07:02.879 ago. That's an impressive jump in efficiency, 00:07:03.139 --> 00:07:05.639 isn't it? It really is. Does that imply we're 00:07:05.639 --> 00:07:08.240 seeing fundamental breakthroughs in AI's learning 00:07:08.240 --> 00:07:10.959 mechanisms? Yeah. Or just better optimization 00:07:10.959 --> 00:07:13.300 of existing models? What do you think? It's likely 00:07:13.300 --> 00:07:15.980 a bit of both, probably. But it certainly shows 00:07:15.980 --> 00:07:18.699 how quickly these models are refining their ability 00:07:18.699 --> 00:07:21.379 to learn complex tasks. Right. And speaking of 00:07:21.379 --> 00:07:23.420 refinement, a crucial point this brings up is 00:07:23.420 --> 00:07:25.699 how AI is beginning to interact with us in more 00:07:25.699 --> 00:07:29.000 sophisticated ways. Anthropic, for example. Claude's 00:07:29.000 --> 00:07:31.980 makers. Exactly. They gave their Claude Opus 00:07:31.980 --> 00:07:34.959 4 and 4 .1 models the ability to actually end 00:07:34.959 --> 00:07:37.600 a conversation in extreme cases. It can literally 00:07:37.600 --> 00:07:40.800 say no and walk away. Huh. That's a big step 00:07:40.800 --> 00:07:44.019 for AI autonomy. It is. Also, Claude introduced 00:07:44.019 --> 00:07:46.779 something called learning mode, which guides 00:07:46.779 --> 00:07:49.399 users to answers rather than just handing them 00:07:49.399 --> 00:07:51.990 over directly. Oh, interesting. How does that 00:07:51.990 --> 00:07:54.290 work? Well, it sort of nudges you, helps you 00:07:54.290 --> 00:07:55.769 figure it out. This isn't just about getting 00:07:55.769 --> 00:07:58.290 information. It begins to define an AI's character, 00:07:58.550 --> 00:08:01.589 you know, by providing a window into its reasoning 00:08:01.589 --> 00:08:03.949 process, a little more transparency into how 00:08:03.949 --> 00:08:05.910 it thinks. That's fascinating. We're also seeing 00:08:05.910 --> 00:08:08.350 new specialized tools emerging, aren't we? Moving 00:08:08.350 --> 00:08:10.949 beyond just general chatbots into very targeted 00:08:10.949 --> 00:08:13.990 applications. Totally. Think of like GPT -5 SEO 00:08:13.990 --> 00:08:16.769 for deeper brand and competitor insights, maybe. 00:08:17.560 --> 00:08:20.420 Or dabe .io, which is a platform designed for 00:08:20.420 --> 00:08:22.860 building custom AI agents just by describing 00:08:22.860 --> 00:08:25.319 the tasks you want them to do. And then there's 00:08:25.319 --> 00:08:28.360 Vydacia for creating jaw -dropping videos, which, 00:08:28.459 --> 00:08:31.759 OK, bold claim. But the progress in video generation 00:08:31.759 --> 00:08:35.059 is undeniable. Even in fashion, this style is 00:08:35.059 --> 00:08:37.600 a new social media app powered by AI suggesting 00:08:37.600 --> 00:08:40.340 outfits and styles. It's all about utility, finding 00:08:40.340 --> 00:08:42.820 a specific need. Exactly. And on the financial 00:08:42.820 --> 00:08:45.600 front, tying back to our first segment, Meta 00:08:45.600 --> 00:08:47.899 isn't just raising debt for its own data centers. 00:08:48.080 --> 00:08:51.320 They're also reportedly raising up to $29 billion 00:08:51.320 --> 00:08:55.059 for broader AI projects, including a massive 00:08:55.059 --> 00:08:59.519 $14 .3 billion investment in scale AI, giving 00:08:59.519 --> 00:09:04.360 them a 49 % stake. 49%. Yeah. It's a clear signal 00:09:04.360 --> 00:09:07.220 of AI's deep integration into financial strategies, 00:09:07.279 --> 00:09:09.720 not just as an expense, but as a strategic investment 00:09:09.720 --> 00:09:11.840 target itself. And we've also got quick hits 00:09:11.840 --> 00:09:15.159 like 11 Labs Jingle Maker, which can turn any 00:09:15.159 --> 00:09:18.419 website into a catchy jingle. Huh. You bet. Perfect 00:09:18.419 --> 00:09:19.960 for when you need a little earworm for your online 00:09:19.960 --> 00:09:22.820 presence, I guess. And NVIDIA. The chip giant 00:09:22.820 --> 00:09:25.899 just launched Granary. Granary. Yeah, a massive 00:09:25.899 --> 00:09:28.299 audio data set with over 1 million hours of audio 00:09:28.299 --> 00:09:31.860 in 25 languages. The sheer variety and specificity 00:09:31.860 --> 00:09:34.379 of progress is truly impressive. It really is. 00:09:34.500 --> 00:09:36.480 It's not just about one breakthrough. It's like 00:09:36.480 --> 00:09:38.960 a thousand small practical innovations happening 00:09:38.960 --> 00:09:41.779 all at once. So thinking about all these diverse, 00:09:41.940 --> 00:09:45.440 specific advancements. What practical impact 00:09:45.440 --> 00:09:47.779 are these new capabilities actually having on 00:09:47.779 --> 00:09:50.259 us, the users, right now? Well, the practical 00:09:50.259 --> 00:09:53.399 impact is becoming profound. AI is getting more 00:09:53.399 --> 00:09:56.659 sophisticated and genuinely helpful in countless 00:09:56.659 --> 00:10:00.039 daily tasks, really, from creative work to strategic 00:10:00.039 --> 00:10:03.610 analysis. Welcome back. We've explored the staggering 00:10:03.610 --> 00:10:06.850 financial investment in AI and touched upon some 00:10:06.850 --> 00:10:09.129 fascinating new capabilities. Now let's talk 00:10:09.129 --> 00:10:11.429 about a big, perhaps unexpected shift in the 00:10:11.429 --> 00:10:14.669 broader AI world. After the release of OpenAI's 00:10:14.669 --> 00:10:17.590 GPT -5, a lot of people are asking, are we in 00:10:17.590 --> 00:10:20.320 a way... hitting an AI wall or maybe just a phase 00:10:20.320 --> 00:10:23.139 of more measured progress. What's truly compelling 00:10:23.139 --> 00:10:25.980 here is the community's reaction. I think GPT 00:10:25.980 --> 00:10:28.779 -5 was without a doubt the most hyped model ever. 00:10:28.919 --> 00:10:30.820 Absolutely. The buzz was huge. Yeah. There were 00:10:30.820 --> 00:10:33.460 whispers, even expectations of jaw dropping demos 00:10:33.460 --> 00:10:35.679 and AGI level breakthroughs, you know, pushing 00:10:35.679 --> 00:10:37.259 the boundaries of what we thought was possible. 00:10:37.440 --> 00:10:39.700 But instead, what people got was something being 00:10:39.700 --> 00:10:42.919 called kind of widely a mid -tier upgrade. A 00:10:42.919 --> 00:10:46.139 mid -tier upgrade. Yeah. It's a bit faster. Yes, 00:10:46.240 --> 00:10:49.559 it's cheaper. And it has better vibes, as some 00:10:49.559 --> 00:10:52.379 put it, meaning it's maybe more reliable or less 00:10:52.379 --> 00:10:55.500 prone to weird errors. OK, better vibes. I like 00:10:55.500 --> 00:10:58.440 that. But even AI insiders were reportedly surprised. 00:10:58.639 --> 00:11:00.799 They expected something truly groundbreaking, 00:11:00.899 --> 00:11:03.360 like a major leap, and they didn't quite get 00:11:03.360 --> 00:11:06.600 that. So it seems GPT -5 is fine. It's useful. 00:11:06.779 --> 00:11:09.559 But maybe don't expect magic, at least not the 00:11:09.559 --> 00:11:11.429 kind of magic some were anticipating. Pretty 00:11:11.429 --> 00:11:13.570 much. And this has fundamentally shifted the 00:11:13.570 --> 00:11:15.809 whole vibe of the AI race, actually. How so? 00:11:16.029 --> 00:11:18.570 Before, the conversation was almost exclusively 00:11:18.570 --> 00:11:21.789 about building AGI, artificial general intelligence, 00:11:21.950 --> 00:11:24.929 right? That elusive human level AI. The holy 00:11:24.929 --> 00:11:27.149 grail. Yeah, that would automate everything and 00:11:27.149 --> 00:11:29.669 revolutionize society overnight. There is this 00:11:29.669 --> 00:11:32.350 pervasive narrative that every model is exponentially 00:11:32.350 --> 00:11:35.210 smarter and will automate everything soon. Very 00:11:35.210 --> 00:11:38.370 future focused, very grandiose. But now the conversation 00:11:38.370 --> 00:11:41.409 is really changing. It is. It's become more grounded. 00:11:41.549 --> 00:11:45.269 The focus is shifting to let's build useful products. 00:11:45.769 --> 00:11:48.649 Practicality. Exactly. There's a growing consensus 00:11:48.649 --> 00:11:51.629 that maybe AGI is further off than we thought, 00:11:51.789 --> 00:11:55.269 which is a significant recalibration. And critically, 00:11:55.470 --> 00:11:58.259 the understanding that... LLM's large language 00:11:58.259 --> 00:12:02.139 models like GPT are infrastructure, not magic. 00:12:02.559 --> 00:12:04.820 Infrastructure, not magic. That's a key phrase. 00:12:04.980 --> 00:12:07.399 They are powerful tools for sure, but they have 00:12:07.399 --> 00:12:09.779 limitations and are part of a larger system. 00:12:10.039 --> 00:12:12.539 I certainly resonate with that shift to utility. 00:12:12.779 --> 00:12:14.820 I mean, I still wrestle with prompt drift myself. 00:12:15.059 --> 00:12:16.879 You know that thing? Oh, yeah, definitely. Where 00:12:16.879 --> 00:12:18.600 you try to get consistent results from an AI, 00:12:18.840 --> 00:12:22.100 but its responses can subtly vary each time you 00:12:22.100 --> 00:12:25.070 prompt it. even with the exact same input. So 00:12:25.070 --> 00:12:28.389 this post -hype focus on making these tools more 00:12:28.389 --> 00:12:31.070 reliable and consistently useful rather than 00:12:31.070 --> 00:12:33.909 chasing elusive breakthroughs. Well, it resonates 00:12:33.909 --> 00:12:36.009 deeply with my own experience using them. And 00:12:36.009 --> 00:12:38.049 if we connect this to the bigger picture, even 00:12:38.049 --> 00:12:40.330 AI regulation has seen a similar shift in tone. 00:12:40.710 --> 00:12:43.190 Under the Biden administration, much of the discussion 00:12:43.190 --> 00:12:45.769 and focus was around, like, existential risk 00:12:45.769 --> 00:12:48.860 and AGI safety. Right. The big what ifs. Yeah. 00:12:48.960 --> 00:12:51.559 Ensuring that super intelligent A .I. wouldn't 00:12:51.559 --> 00:12:54.720 harm humanity. But our sources highlight that 00:12:54.720 --> 00:12:57.500 under a potential Trump administration, the mood 00:12:57.500 --> 00:12:59.759 around A .I. regulation has reportedly become 00:12:59.759 --> 00:13:03.419 way more chill. Focusing less on those speculative 00:13:03.419 --> 00:13:07.000 risks and more on economic competitiveness and 00:13:07.000 --> 00:13:09.860 just getting it deployed. This political shift 00:13:09.860 --> 00:13:12.240 truly reflects the broader recalibration we're 00:13:12.240 --> 00:13:14.490 seeing across the industry. It feels like we're 00:13:14.490 --> 00:13:17.090 genuinely entering the post -type era of AI, 00:13:17.309 --> 00:13:19.889 doesn't it? A phase where even the smartest people 00:13:19.889 --> 00:13:23.330 in the room are... recalibrating their goals, 00:13:23.350 --> 00:13:25.309 their expectations. Yeah, I think so. It's less 00:13:25.309 --> 00:13:28.029 about the theoretical singularity and more about 00:13:28.029 --> 00:13:30.169 solving tangible problems. You can almost call 00:13:30.169 --> 00:13:32.190 it a maturity phase for the technology. A maturity 00:13:32.190 --> 00:13:35.250 phase. That sounds right. So for us, the users, 00:13:35.409 --> 00:13:37.190 the businesses, anyone interacting with this 00:13:37.190 --> 00:13:39.950 technology, what does this post -hype era truly 00:13:39.950 --> 00:13:43.350 mean for how we approach AI? Well, I think this 00:13:43.350 --> 00:13:47.090 post -hype era means a renewed focus on... Practical 00:13:47.090 --> 00:13:50.250 applications, tangible value, and maybe more 00:13:50.250 --> 00:13:53.450 realistic expectations for AI's current capabilities. 00:13:53.970 --> 00:13:55.929 So what does this all mean when we put it all 00:13:55.929 --> 00:13:59.490 together? We've really seen this staggering debt 00:13:59.490 --> 00:14:02.509 -fueled financial investment in AI infrastructure, 00:14:02.710 --> 00:14:05.669 which is building the very foundation of this 00:14:05.669 --> 00:14:08.629 new technological era. It's an enormous financial 00:14:08.629 --> 00:14:10.970 gamble, really, one that could reshape global 00:14:10.970 --> 00:14:13.769 economics. And simultaneously, we're seeing a 00:14:13.769 --> 00:14:15.950 really fascinating pivot within the AI community 00:14:15.950 --> 00:14:18.929 itself, a shift from that grand, sometimes unrealistic 00:14:18.929 --> 00:14:22.570 hype of achieving AGI towards a more grounded 00:14:22.570 --> 00:14:25.549 pursuit of practical utility. It's a clear signal 00:14:25.549 --> 00:14:28.269 that AI's biggest risks might truly be financial 00:14:28.269 --> 00:14:31.809 and systemic rather than just technical or existential. 00:14:32.350 --> 00:14:35.289 It seems AI isn't slowing down. It's simply evolving 00:14:35.289 --> 00:14:37.929 its focus. Right. Becoming more grounded, more 00:14:37.929 --> 00:14:40.509 about building genuinely useful tools and solving 00:14:40.509 --> 00:14:42.740 real -world problems. problems than chasing those 00:14:42.740 --> 00:14:45.059 distant, often sci -fi -inspired dreams. It's 00:14:45.059 --> 00:14:47.360 an exciting, albeit maybe more pragmatic phase. 00:14:47.559 --> 00:14:50.039 Thank you for joining us on this deep dive into 00:14:50.039 --> 00:14:53.399 the fascinating world of AI's present reality 00:14:53.399 --> 00:14:56.960 and its evolving future. And here's a provocative 00:14:56.960 --> 00:15:00.480 thought for you to mull over. If AI is now firmly 00:15:00.480 --> 00:15:02.580 established as a foundational infrastructure, 00:15:03.019 --> 00:15:06.259 much like electricity or the internet, How does 00:15:06.259 --> 00:15:08.620 that fundamentally change who controls it? Ooh, 00:15:08.659 --> 00:15:11.039 good question. And maybe more importantly, what 00:15:11.039 --> 00:15:13.440 new accessible innovations might that enable 00:15:13.440 --> 00:15:16.200 for everyone, not just the tech giants? Something 00:15:16.200 --> 00:15:17.940 to think about. Keep learning, keep exploring. 00:15:18.220 --> 00:15:19.919 We'll see you next time on The Deep Dive.