WEBVTT 00:00:00.000 --> 00:00:03.660 Imagine a massive warehouse for a second. Beat, 00:00:03.839 --> 00:00:06.660 concrete and steel stretching out endlessly into 00:00:06.660 --> 00:00:09.140 the horizon. Just massive physical scale. Right. 00:00:09.240 --> 00:00:12.400 And inside this structure, tech giants are pouring 00:00:12.400 --> 00:00:15.240 a trillion dollars. You hear the deafening roar 00:00:15.240 --> 00:00:18.399 of raw compute power. Yeah, it's a literal physical 00:00:18.399 --> 00:00:22.300 monument to human ambition. It really is. But 00:00:22.300 --> 00:00:25.600 then I want you to picture. A quiet, sterile 00:00:25.600 --> 00:00:28.699 hospital room. An invisible piece of code reviews 00:00:28.699 --> 00:00:32.060 a standard medical scan. It spots something completely 00:00:32.060 --> 00:00:35.140 imperceptible to human eyes. And it silently 00:00:35.140 --> 00:00:38.539 saves human life. Wow. That striking contrast 00:00:38.539 --> 00:00:41.060 is exactly what we are unpacking today. Yeah. 00:00:41.159 --> 00:00:43.740 We're looking at the sheer scale of global hardware, 00:00:43.880 --> 00:00:45.719 and we're connecting it to those deeply intimate 00:00:45.719 --> 00:00:48.100 human impacts. The journey between those two 00:00:48.100 --> 00:00:50.899 extremes is absolutely fascinating to Sex Silence. 00:00:51.100 --> 00:00:53.200 Welcome to today's deep dive. We are thrilled 00:00:53.200 --> 00:00:55.000 you are joining us for this conversation. We 00:00:55.000 --> 00:00:56.979 really are. We're skipping the usual intro housekeeping 00:00:56.979 --> 00:00:59.520 today. Yeah, the stakes of this technological 00:00:59.520 --> 00:01:03.460 shift are just too high. So here is our roadmap 00:01:03.460 --> 00:01:06.129 for today. First, we're following the massive 00:01:06.129 --> 00:01:09.310 money trail. We'll analyze big tech's trillion 00:01:09.310 --> 00:01:12.310 -dollar hardware race. Following the money is 00:01:12.310 --> 00:01:15.530 always step one. Always. Second, we'll see how 00:01:15.530 --> 00:01:17.629 that money births highly specialized software 00:01:17.629 --> 00:01:20.250 tools. Finally, we'll examine a breathtaking 00:01:20.250 --> 00:01:23.170 medical breakthrough. The Mayo Clinic is proving 00:01:23.170 --> 00:01:26.189 this tech fundamentally alters human survival. 00:01:26.799 --> 00:01:29.659 It's a profound shift. We are moving from raw 00:01:29.659 --> 00:01:32.400 infrastructure to an actual guardian. Right. 00:01:32.519 --> 00:01:34.780 The scale of the foundational investment is just 00:01:34.780 --> 00:01:37.120 hard to comprehend. Let's start with the sheer 00:01:37.120 --> 00:01:39.400 magnitude of that spending. The numbers are wild. 00:01:39.579 --> 00:01:41.700 They really are. Wall Street analysts are projecting 00:01:41.700 --> 00:01:44.659 something truly staggering right now. Big tech's 00:01:44.659 --> 00:01:47.959 AI spending will hit $1 trillion soon. $1 trillion. 00:01:48.120 --> 00:01:50.579 Specifically, they expect to cross that threshold 00:01:50.579 --> 00:01:54.359 by 2027. Two sec silence. That is roughly the 00:01:54.359 --> 00:01:56.519 entire gross domestic product of the Netherlands. 00:01:56.780 --> 00:01:59.200 Yeah, an entire developed nation's economic output 00:01:59.200 --> 00:02:02.560 in a single year. And it is all being poured 00:02:02.560 --> 00:02:05.180 directly into data centers. They're just buying 00:02:05.180 --> 00:02:08.400 up chips, servers, and massive cooling infrastructure. 00:02:08.819 --> 00:02:11.979 Amazon alone is spending $200 billion on this. 00:02:12.180 --> 00:02:16.889 $200 billion. Just Amazon. Right. Andy Jassy 00:02:16.889 --> 00:02:19.650 is aggressively doubling down on their AWS build 00:02:19.650 --> 00:02:23.330 out, but they're also heavily investing in custom 00:02:23.330 --> 00:02:25.629 silicon. Which is super interesting. Yeah, they're 00:02:25.629 --> 00:02:28.110 pushing their own specialized tranium chips. 00:02:28.349 --> 00:02:31.030 That represents a massive strategic pivot. They 00:02:31.030 --> 00:02:33.310 really want to control their own destiny here. 00:02:33.469 --> 00:02:35.789 Right. They want to drastically reduce reliance 00:02:35.789 --> 00:02:39.060 on outside hardware vendors. Exactly. Building 00:02:39.060 --> 00:02:41.879 custom silicon gives them crucial leverage in 00:02:41.879 --> 00:02:44.020 the market. It's smart. Then you have Microsoft 00:02:44.020 --> 00:02:47.680 spending $190 billion. Also a staggering figure. 00:02:47.879 --> 00:02:50.379 Their capital expenditure is up a whopping 24 00:02:50.379 --> 00:02:54.060 % lately. They are racing to keep Azure AI and 00:02:54.060 --> 00:02:57.020 Copilot ahead globally. Right, and Alphabet is 00:02:57.020 --> 00:03:00.300 right behind them at $185 billion. Yeah. Google's 00:03:00.300 --> 00:03:03.300 cloud business is actually growing at 63 % year 00:03:03.300 --> 00:03:05.840 over year. That is blistering growth. It really 00:03:05.840 --> 00:03:09.479 is. That kind of demand requires a massive infrastructure 00:03:09.479 --> 00:03:12.500 scale up. You just cannot support those customers 00:03:12.500 --> 00:03:15.539 without building new data centers. And Meta is 00:03:15.539 --> 00:03:20.460 sitting at $135 billion. That spending is mostly 00:03:20.460 --> 00:03:23.699 driven by rising memory component costs. And 00:03:23.699 --> 00:03:27.080 they're quietly building out their massive superintelligence 00:03:27.080 --> 00:03:29.969 lab. But I have to push back on this narrative 00:03:29.969 --> 00:03:32.729 a little bit. Okay, let's hear it. This level 00:03:32.729 --> 00:03:35.830 of capital expenditure honestly feels completely 00:03:35.830 --> 00:03:39.210 reckless to me. Really? Reckless? Yeah, it's 00:03:39.210 --> 00:03:41.069 like laying down continent -spanning train tracks 00:03:41.069 --> 00:03:43.169 right now. But we're laying them before we even 00:03:43.169 --> 00:03:45.949 know where the cargo is going. Ah. Well, that 00:03:45.949 --> 00:03:48.689 train track analogy captures this hardware phase 00:03:48.689 --> 00:03:51.150 perfectly. You think so? Definitely. Yeah. You 00:03:51.150 --> 00:03:53.270 absolutely have to lay the steel before the trains 00:03:53.270 --> 00:03:56.050 can run. If you wait for the actual cargo, you've 00:03:56.050 --> 00:03:58.400 already lost the race. I see your point. But 00:03:58.400 --> 00:04:00.439 the actual type of steel we need is fundamentally 00:04:00.439 --> 00:04:03.099 changing. I really want to dig into that specific 00:04:03.099 --> 00:04:05.650 hardware shift. Let's do it. For the last couple 00:04:05.650 --> 00:04:09.090 of years, it has been all about GPUs. Graphics 00:04:09.090 --> 00:04:11.949 processing units have been the absolute undisputed 00:04:11.949 --> 00:04:15.550 kings. Nvidia has completely dominated this entire 00:04:15.550 --> 00:04:18.209 technological cycle. Right. GPUs are fantastic 00:04:18.209 --> 00:04:20.529 for the brute force of training large models. 00:04:20.670 --> 00:04:23.110 They handle thousands of simple math problems 00:04:23.110 --> 00:04:26.730 simultaneously. But analysts are starting to 00:04:26.730 --> 00:04:30.589 predict a massive CPU renaissance. Central processing 00:04:30.589 --> 00:04:33.329 units are stepping back into the spotlight. Yeah. 00:04:33.329 --> 00:04:35.589 Why are we going... Going back to standard CPUs 00:04:35.589 --> 00:04:37.089 all of a sudden, that feels like a technological 00:04:37.089 --> 00:04:40.069 step backward on the surface. It kind of does, 00:04:40.089 --> 00:04:42.629 right? Yeah. But it's because the software is 00:04:42.629 --> 00:04:45.310 fundamentally changing underneath us. We are 00:04:45.310 --> 00:04:48.430 moving heavily into the era of agentic AI. We 00:04:48.430 --> 00:04:50.639 should define that. turn clearly before we go 00:04:50.639 --> 00:04:52.920 any further. What does agentic AI actually mean 00:04:52.920 --> 00:04:55.720 in this specific hardware context? Sure. It's 00:04:55.720 --> 00:04:59.019 AI that independently finishes tasks without 00:04:59.019 --> 00:05:01.720 needing constant human input. So it's not just 00:05:01.720 --> 00:05:03.920 waiting for you to type a prompt. Exactly. It's 00:05:03.920 --> 00:05:06.259 operating autonomously in the background. Right. 00:05:06.889 --> 00:05:09.589 It operates in these complex, long -running sequential 00:05:09.589 --> 00:05:13.269 workflows. It has to verify step A before it 00:05:13.269 --> 00:05:16.509 can even attempt step B. Ah, I see. And traditional 00:05:16.509 --> 00:05:19.430 CPUs are actually much better at handling sequential 00:05:19.430 --> 00:05:22.589 logic tasks. Because GPUs are great for parallel 00:05:22.589 --> 00:05:25.750 processing. Yes, exactly. But agents require 00:05:25.750 --> 00:05:29.160 deep logical steps. One after the other. So this 00:05:29.160 --> 00:05:31.399 represents incredible news for older companies 00:05:31.399 --> 00:05:33.959 like Intel and Arm. Absolutely. They provide 00:05:33.959 --> 00:05:36.660 the fundamental backbone for these complex logic 00:05:36.660 --> 00:05:40.579 processes. The hardware landscape is diversifying 00:05:40.579 --> 00:05:43.139 to match the software evolution. We're moving 00:05:43.139 --> 00:05:45.480 from brute force training to complex logical 00:05:45.480 --> 00:05:48.240 execution. Which brings up a really crucial question 00:05:48.240 --> 00:05:50.899 about all this money. How is Wall Street reacting 00:05:50.899 --> 00:05:53.240 to these companies essentially burning cash for 00:05:53.240 --> 00:05:55.600 future capabilities? Well, Wall Street is definitely 00:05:55.600 --> 00:05:57.759 anxious. They worry about short term profit margins 00:05:57.759 --> 00:05:59.959 taking a massive hit. Not true. Investors always 00:05:59.959 --> 00:06:02.060 want to see immediate returns on capital expenditures. 00:06:02.500 --> 00:06:05.319 But the tech executives view this as a long term 00:06:05.319 --> 00:06:08.100 survival strategy. Right. If you miss this infrastructure 00:06:08.100 --> 00:06:10.680 wave, your company becomes entirely obsolete 00:06:10.680 --> 00:06:13.149 tomorrow. So it's less about immediate profit 00:06:13.149 --> 00:06:16.290 and more about corporate survival. Exactly. It's 00:06:16.290 --> 00:06:19.769 a literal existential arms race for the future 00:06:19.769 --> 00:06:21.949 of global computing. That brings us perfectly 00:06:21.949 --> 00:06:24.230 into the second part of our journey today. Yeah, 00:06:24.269 --> 00:06:26.629 the software side. Because this trillion dollar 00:06:26.629 --> 00:06:29.529 hardware foundation is not just sitting idle 00:06:29.529 --> 00:06:33.189 in dark warehouses. It is actively spawning a 00:06:33.189 --> 00:06:35.850 rapid evolution in software capabilities right 00:06:35.850 --> 00:06:38.529 now. We are watching the trains finally run on 00:06:38.529 --> 00:06:40.709 those expensive tracks. We're witnessing a massive 00:06:40.709 --> 00:06:44.329 shift away from simple chat interfaces. AI is 00:06:44.329 --> 00:06:47.209 becoming a proactive collaborator in highly specialized 00:06:47.209 --> 00:06:50.050 digital ecosystems. Yeah, we see it in general 00:06:50.050 --> 00:06:52.949 workflows everywhere you look today. Like Google 00:06:52.949 --> 00:06:55.920 Translate just hit its 20 year end. 20 years. 00:06:56.279 --> 00:06:58.540 That's wild. I know. And they quietly dropped 00:06:58.540 --> 00:07:01.720 20 hidden AI features to celebrate the milestone. 00:07:01.920 --> 00:07:04.579 And Gemini can now export real formatted files, 00:07:04.740 --> 00:07:07.459 not just plain text. You can instantly push out 00:07:07.459 --> 00:07:11.079 docs, sheets, Word and Excel files, mark down 00:07:11.079 --> 00:07:13.439 two. The AI is integrating directly into the 00:07:13.439 --> 00:07:16.120 tools we already use daily. But the hyper specialization 00:07:16.120 --> 00:07:18.579 is what really catches my attention lately. Oh, 00:07:18.600 --> 00:07:20.399 for sure. Let's examine the coding and security 00:07:20.399 --> 00:07:23.860 space for a moment. Mistral AI just dropped their 00:07:23.860 --> 00:07:27.839 new Medium 3 .5 model. Right. And that model 00:07:27.839 --> 00:07:32.560 features a massive 256 ,000 context window. Which 00:07:32.560 --> 00:07:35.839 is huge. It is. To put that in perspective, it 00:07:35.839 --> 00:07:39.540 holds a massive textbook of data in short -term 00:07:39.540 --> 00:07:41.680 memory. It can look at your entire code base 00:07:41.680 --> 00:07:45.060 all at once. Plus, they released these new Vibe 00:07:45.060 --> 00:07:47.439 coding agents to the public. Yeah. They can actually 00:07:47.439 --> 00:07:50.500 ship. GitHub pull requests asynchronously in 00:07:50.500 --> 00:07:53.000 the background. Which fundamentally shifts the 00:07:53.000 --> 00:07:54.899 daily paradigm for software engineers. It really 00:07:54.899 --> 00:07:56.899 does. The AI is doing the actual engineering 00:07:56.899 --> 00:07:59.420 work while you sleep. You wake up, and the code 00:07:59.420 --> 00:08:01.500 is written, tested, and waiting for approval. 00:08:01.720 --> 00:08:03.439 You know, I have a vulnerable admission to make 00:08:03.439 --> 00:08:06.540 right here. I still wrestle with just organizing 00:08:06.540 --> 00:08:08.660 my own basic prompts sometimes. Oh, I think a 00:08:08.660 --> 00:08:10.600 lot of people do. The flood of these autonomous 00:08:10.600 --> 00:08:13.660 features feels a little bit overwhelming. That 00:08:13.660 --> 00:08:16.569 is a completely normal reaction. The blistering 00:08:16.569 --> 00:08:19.649 pace of change is intense. The tech ecosystem 00:08:19.649 --> 00:08:22.550 is fragmenting into highly specific verticals 00:08:22.550 --> 00:08:25.089 very rapidly. We are also seeing this deeply 00:08:25.089 --> 00:08:28.529 impact the cybersecurity sector. Anthropic launched 00:08:28.529 --> 00:08:31.810 the cloud security beta recently for enterprise 00:08:31.810 --> 00:08:34.570 customers. It lets cloud scan entire code bases 00:08:34.570 --> 00:08:37.659 and check if bugs are actually real. It can even 00:08:37.659 --> 00:08:40.259 suggest secure patches before the code hits production 00:08:40.259 --> 00:08:43.659 servers. OpenAI is pushing the envelope on advanced 00:08:43.659 --> 00:08:46.159 security protocols as well. They recently launched 00:08:46.159 --> 00:08:48.779 advanced security mode for chat GPT and codex 00:08:48.779 --> 00:08:51.559 users. Yeah, that update is fascinating. This 00:08:51.559 --> 00:08:53.980 specific feature is actually a little bit terrifying 00:08:53.980 --> 00:08:56.779 to me. Why is that? It removes passwords and 00:08:56.779 --> 00:08:58.659 support recovery entirely from your personal 00:08:58.659 --> 00:09:00.919 account. Oh, right. If you lose your security 00:09:00.919 --> 00:09:03.980 key, you are completely on your own. Your data 00:09:03.980 --> 00:09:06.480 is locked away forever without any... corporate 00:09:06.480 --> 00:09:09.980 backdoor access it is an incredibly high stakes 00:09:09.980 --> 00:09:12.940 approach yeah especially for developers handling 00:09:12.940 --> 00:09:16.399 sensitive code but the tools are just becoming 00:09:16.399 --> 00:09:19.080 incredibly specialized across the board we're 00:09:19.080 --> 00:09:21.580 seeing brand new specialized tools popping up 00:09:21.580 --> 00:09:24.879 for everything imaginable yeah like video s right 00:09:24.879 --> 00:09:27.500 it's now an all -in -one platform for complex 00:09:27.500 --> 00:09:30.990 video marketing workflows And Hera Launch creates 00:09:30.990 --> 00:09:34.409 studio -quality launch videos. It even decides 00:09:34.409 --> 00:09:37.049 the editing pacing automatically. Mintlify Editor 00:09:37.049 --> 00:09:39.950 is offering live, collaborative, Git -synced 00:09:39.950 --> 00:09:42.409 documentation for massive engineering teams. 00:09:42.610 --> 00:09:45.649 That's a game changer. The AI agents update the 00:09:45.649 --> 00:09:47.850 technical docs automatically when someone pushes 00:09:47.850 --> 00:09:50.389 new code. And Wonder puts an autonomous AI design 00:09:50.389 --> 00:09:53.190 agent directly on your creative canvas. It actually 00:09:53.190 --> 00:09:55.669 designs alongside you in real time. It's wild. 00:09:55.870 --> 00:09:58.230 We also have to acknowledge the geopolitical 00:09:58.230 --> 00:10:00.990 reality of these global ecosystems. Yeah, that's 00:10:00.990 --> 00:10:03.110 a crucial layer here. The global supply chain 00:10:03.110 --> 00:10:05.570 of artificial intelligence is incredibly complex 00:10:05.570 --> 00:10:09.019 today. Two U .S. House committees recently opened 00:10:09.019 --> 00:10:11.580 formal probes into specific tech tools. Right. 00:10:11.659 --> 00:10:14.580 They are specifically looking at AnySphere, the 00:10:14.580 --> 00:10:18.259 maker of Cursor, and Airbnb. The probes are focused 00:10:18.259 --> 00:10:21.080 on their use of foreign AI models. Specifically, 00:10:21.279 --> 00:10:23.320 they're examining the integration of models tied 00:10:23.320 --> 00:10:25.879 to Kimi and Quinn. Right. The committees are 00:10:25.879 --> 00:10:28.279 examining international data flows and overall 00:10:28.279 --> 00:10:31.480 security protocols. And we want to be clear here, 00:10:31.539 --> 00:10:33.580 it is very important to note this is strictly 00:10:33.580 --> 00:10:36.620 a fact -finding inquiry. Absolutely. We're maintaining 00:10:36.620 --> 00:10:38.879 absolute neutrality here. We're just reporting 00:10:38.879 --> 00:10:40.980 the facts of the probe without endorsing any 00:10:40.980 --> 00:10:43.879 political viewpoint. Exactly. We are just observing 00:10:43.879 --> 00:10:46.940 how globally intertwined these specific technologies 00:10:46.940 --> 00:10:50.629 have become lately. The geographic origin of 00:10:50.629 --> 00:10:52.950 the intelligence layer is becoming a major global 00:10:52.950 --> 00:10:56.149 focus, but the specialization is also moving 00:10:56.149 --> 00:10:58.789 far beyond just software engineering. It's not 00:10:58.789 --> 00:11:00.970 just about writing emails or generating code 00:11:00.970 --> 00:11:04.830 anymore. For example, Kaikaku AI says food artificial 00:11:04.830 --> 00:11:08.049 intelligence just hit a massive milestone. They're 00:11:08.049 --> 00:11:10.169 calling it a chat GPT moment for the culinary 00:11:10.169 --> 00:11:13.190 world. Their Epicure model is absolutely fascinating 00:11:13.190 --> 00:11:15.750 from a specialized data perspective. It really 00:11:15.750 --> 00:11:19.419 is. It actively learns taste. texture and cuisine 00:11:19.419 --> 00:11:22.580 styles from reading text recipes alone. It translates 00:11:22.580 --> 00:11:25.039 written words into physical flavor profiles. 00:11:25.379 --> 00:11:27.519 And now it's directly powering robotic kitchens 00:11:27.519 --> 00:11:31.220 and automated menu design. We are literally digitizing 00:11:31.220 --> 00:11:33.679 the human experience of taste. Another major 00:11:33.679 --> 00:11:37.019 focus is applying this deep intelligence to complex 00:11:37.019 --> 00:11:39.659 human biology. This is where it gets really profound. 00:11:40.139 --> 00:11:43.100 Mark Zuckerberg's Biohub is investing $500 million 00:11:43.100 --> 00:11:46.299 in AI biology. Yeah, they're putting $400 million 00:11:46.299 --> 00:11:48.940 specifically into massive data tech development. 00:11:50.120 --> 00:11:52.580 And another 100 million is going directly into 00:11:52.580 --> 00:11:54.940 building brand new laboratory spaces. They want 00:11:54.940 --> 00:11:57.740 to build massive cellular data sets to train 00:11:57.740 --> 00:12:00.480 entirely new predictive models. The ultimate 00:12:00.480 --> 00:12:03.019 goal is to accurately predict disease behavior 00:12:03.019 --> 00:12:05.919 before it ever happens. Which is the holy grail 00:12:05.919 --> 00:12:08.299 of specialized artificial intelligence applications. 00:12:08.799 --> 00:12:11.240 Truly. But looking at all these different autonomous 00:12:11.240 --> 00:12:13.379 agents, I have to ask a question. Sure. Go ahead. 00:12:13.799 --> 00:12:17.639 Does this hyper specialization mean humans will 00:12:17.639 --> 00:12:21.559 just become managers of very... Well, the nature 00:12:21.559 --> 00:12:23.159 of our daily professional work is definitely 00:12:23.159 --> 00:12:26.299 going to pivot. Human oversight will shift away 00:12:26.299 --> 00:12:29.179 from doing the raw, manual, repetitive work. 00:12:29.379 --> 00:12:32.179 We'll be focused entirely on directing the overarching 00:12:32.179 --> 00:12:34.759 strategic workflow of these systems. Right. We 00:12:34.759 --> 00:12:36.600 stop being the mechanics and start being the 00:12:36.600 --> 00:12:39.960 air traffic controllers. Exactly. You guide the 00:12:39.960 --> 00:12:41.980 autonomous agents instead of turning the heavy 00:12:41.980 --> 00:12:45.299 wrenches yourself. You manage the airspace while 00:12:45.299 --> 00:12:48.309 the machines fly the individual planes. Let's 00:12:48.309 --> 00:12:50.490 take a brief pause here before our final topic, 00:12:50.690 --> 00:12:56.009 sponsor. We are back. Beat. That massive biological 00:12:56.009 --> 00:12:58.889 investment from Zuckerberg is largely about the 00:12:58.889 --> 00:13:01.330 distant future. Right. It's laying the groundwork. 00:13:01.429 --> 00:13:04.090 But this incredible technology is already operating 00:13:04.090 --> 00:13:07.070 in human hospitals today. All this raw computing 00:13:07.070 --> 00:13:09.590 power is actually culminating in saving real 00:13:09.590 --> 00:13:12.490 human lives. This is where the hardware and software 00:13:12.490 --> 00:13:15.620 become. deeply personal for everyone. It stops 00:13:15.620 --> 00:13:18.519 being abstract technology and becomes a deeply 00:13:18.519 --> 00:13:21.379 human story. Let us examine the Mayo Clinic's 00:13:21.379 --> 00:13:23.679 incredible new AI breakthrough right now. They 00:13:23.679 --> 00:13:25.899 have officially deployed a new diagnostic AI 00:13:25.899 --> 00:13:29.279 system called Red Mode. It is specifically designed 00:13:29.279 --> 00:13:31.879 to detect the early signs of pancreatic cancer. 00:13:32.350 --> 00:13:34.529 And pancreatic cancer is historically one of 00:13:34.529 --> 00:13:37.769 the absolute toughest diseases we face. The five 00:13:37.769 --> 00:13:39.850 -year survival rate is currently sitting tragically 00:13:39.850 --> 00:13:42.870 below 15 % globally. And that devastatingly low 00:13:42.870 --> 00:13:45.210 survival rate is mainly due to late detection. 00:13:45.509 --> 00:13:48.210 By the time physical symptoms show up, the disease 00:13:48.210 --> 00:13:50.750 has usually spread everywhere. But Red Mode is 00:13:50.750 --> 00:13:53.029 actually reaching back in time to spot the silent 00:13:53.029 --> 00:13:57.269 cancer. It finds the subtle warning signs years 00:13:57.269 --> 00:13:59.769 before it becomes a clinical crisis. The empirical 00:13:59.769 --> 00:14:02.309 data on this system is just completely mind -blowing 00:14:02.309 --> 00:14:04.809 to me. It's unbelievable. The researchers had 00:14:04.809 --> 00:14:08.289 Red Mode look at nearly 2 ,000 historical medical 00:14:08.289 --> 00:14:11.669 scans. These were older routine scans that were 00:14:11.669 --> 00:14:14.309 originally cleared by human medical specialists. 00:14:14.730 --> 00:14:17.250 Meaning highly trained doctors looked at these 00:14:17.250 --> 00:14:19.610 exact images and saw absolutely nothing wrong. 00:14:19.809 --> 00:14:22.409 Right. The scans were deemed completely normal 00:14:22.409 --> 00:14:25.909 by top human experts. Exactly. But the AI successfully 00:14:25.909 --> 00:14:29.730 identified 73 % of those hidden cancer cases. 00:14:29.929 --> 00:14:32.549 Wow. And it found them up to three years before 00:14:32.549 --> 00:14:35.269 the actual clinical diagnosis happened. Three 00:14:35.269 --> 00:14:37.909 years is an absolute eternity in the high -stakes 00:14:37.909 --> 00:14:40.169 world of oncology. It's the crucial difference 00:14:40.169 --> 00:14:42.629 between proactive treatment and palliative care. 00:14:42.809 --> 00:14:45.669 At the two -year mark, the AI diagnostic results 00:14:45.669 --> 00:14:48.350 are even more staggering. The system is three 00:14:48.350 --> 00:14:50.590 times more effective at spotting early -stage 00:14:50.590 --> 00:14:53.200 cancer anomalies. Three times. That is compared 00:14:53.200 --> 00:14:56.139 to even the most experienced, highly trained 00:14:56.139 --> 00:14:59.240 human radiologists. It's incredibly important 00:14:59.240 --> 00:15:01.220 to understand how it's actually achieving this 00:15:01.220 --> 00:15:04.240 feat. The mathematical model analyzes hundreds 00:15:04.240 --> 00:15:06.740 of different quantitative features hidden in 00:15:06.740 --> 00:15:09.399 the scan. These are incredibly subtle textural 00:15:09.399 --> 00:15:12.460 changes in the biological tissue itself. Yeah, 00:15:12.500 --> 00:15:15.039 features that are physically impossible for a 00:15:15.039 --> 00:15:17.960 human brain to process visually. The human eye 00:15:17.960 --> 00:15:21.080 biologically cannot perceive the mathematical 00:15:21.080 --> 00:15:25.240 pixel patterns it's finding. It's seeing a reality 00:15:25.240 --> 00:15:27.139 that we are blind to. And what makes this so 00:15:27.139 --> 00:15:29.500 revolutionary is the actual deployment method 00:15:29.500 --> 00:15:33.059 used. It runs entirely on standard CT scans that 00:15:33.059 --> 00:15:35.399 patients already get every single day. Which 00:15:35.399 --> 00:15:37.279 means it can be layered directly into routine 00:15:37.279 --> 00:15:40.029 medical care instantly. Yes. It adds zero extra 00:15:40.029 --> 00:15:42.429 financial costs or diagnostic friction for the 00:15:42.429 --> 00:15:44.690 hospital patient. You do not need a special machine 00:15:44.690 --> 00:15:46.769 or an invasive new procedure. It essentially 00:15:46.769 --> 00:15:49.389 moves the entire medical diagnosis window up 00:15:49.389 --> 00:15:52.769 by 36 months. We often talk about AI purely as 00:15:52.769 --> 00:15:54.889 a productivity tool for office work. You know, 00:15:54.909 --> 00:15:58.570 we focus on emails, cogeneration, and financial 00:15:58.570 --> 00:16:01.750 efficiency. But in this specific case, it's acting 00:16:01.750 --> 00:16:06.450 as a silent, vigilant guardian. Whoa. Beat. Imagine 00:16:06.450 --> 00:16:09.259 an AI. Quietly saving your life in the background 00:16:09.259 --> 00:16:12.000 while just checking a sore back. Two sec silence. 00:16:12.639 --> 00:16:15.450 You go in for a routine checkup. And a machine 00:16:15.450 --> 00:16:18.169 alters your destiny. That is the exact reality 00:16:18.169 --> 00:16:20.490 the Mayo Clinic is actively building today. Yeah. 00:16:20.570 --> 00:16:23.830 It turns every single routine medical visit into 00:16:23.830 --> 00:16:26.450 a potential life -saving screening event. It 00:16:26.450 --> 00:16:28.429 really makes you pause and deeply consider the 00:16:28.429 --> 00:16:30.990 implications of this shift. It does. Does this 00:16:30.990 --> 00:16:33.730 kind of hidden superhuman pattern matching make 00:16:33.730 --> 00:16:37.070 the human doctor entirely obsolete? If a machine 00:16:37.070 --> 00:16:39.490 sees what we cannot, where does the human fit 00:16:39.490 --> 00:16:42.039 in? The human doctor absolutely does not become 00:16:42.039 --> 00:16:45.039 obsolete in this new medical paradigm. Doctors 00:16:45.039 --> 00:16:47.299 will remain the deeply empathic, critical decision 00:16:47.299 --> 00:16:49.919 makers in the exam room. They'll simply be armed 00:16:49.919 --> 00:16:52.159 with superhuman diagnostic foresight from the 00:16:52.159 --> 00:16:55.000 AI systems. Medicine is ultimately about human 00:16:55.000 --> 00:16:56.940 connection and guiding people through terrifying 00:16:56.940 --> 00:16:59.340 choices. They become the ultimate decision makers 00:16:59.340 --> 00:17:01.519 armed with real -life, time -traveling insights. 00:17:01.840 --> 00:17:05.079 That captures the dynamic perfectly. The human 00:17:05.079 --> 00:17:07.460 emotional element becomes even more important 00:17:07.460 --> 00:17:10.039 when delivering that complex medical news. Right. 00:17:10.160 --> 00:17:12.859 The machine finds the invisible threat, but the 00:17:12.859 --> 00:17:15.599 human doctor treats the actual patient. We have 00:17:15.599 --> 00:17:18.039 covered a truly massive amount of ground today. 00:17:18.220 --> 00:17:20.900 We really have explored a vast landscape. If 00:17:20.900 --> 00:17:23.779 we connect this entire journey together, it paints 00:17:23.779 --> 00:17:26.599 a pretty profound picture. We started with massive. 00:17:27.240 --> 00:17:30.500 concrete server farms consuming the GDP of small 00:17:30.500 --> 00:17:33.619 European nations. We watched that brute force 00:17:33.619 --> 00:17:36.700 compute power trickle down into highly specialized 00:17:36.700 --> 00:17:39.599 software agents. And it ultimately culminates 00:17:39.599 --> 00:17:42.839 in a silent algorithm guardian that actively 00:17:42.839 --> 00:17:45.819 rewrites human longevity. It is the trillion 00:17:45.819 --> 00:17:48.200 dollar hardware receipt finally paying out in 00:17:48.200 --> 00:17:50.599 extended human years. The overarching return 00:17:50.599 --> 00:17:52.599 on investment isn't just measured in financial 00:17:52.599 --> 00:17:55.180 profit margins. It's fundamentally measured in 00:17:55.180 --> 00:17:57.619 human time. I want to leave you with a final 00:17:57.619 --> 00:18:00.220 lingering thought for today. If an artificial 00:18:00.220 --> 00:18:02.539 intelligence can spot the invisible signs of 00:18:02.539 --> 00:18:05.539 a fatal disease three years early just by looking 00:18:05.539 --> 00:18:08.259 at a standard medical scan, what other silent 00:18:08.259 --> 00:18:10.440 patterns is it already seeing in the background 00:18:10.440 --> 00:18:12.359 of our daily data that we haven't even thought 00:18:12.359 --> 00:18:15.539 to ask about yet to sex silence? Thank you so 00:18:15.539 --> 00:18:17.339 much for taking this deep dive with us today. 00:18:17.380 --> 00:18:19.799 We will catch you on the next one. Out Tiro music.