1 00:00:00,080 --> 00:00:03,439 Speaker 1: I want you to imagine waking up tomorrow morning in 2 00:00:03,480 --> 00:00:08,519 a world that feels well, it feels distinctly different but 3 00:00:08,599 --> 00:00:12,080 also structurally familiar. You know, you step out of bed 4 00:00:12,359 --> 00:00:15,439 and the house around you is quietly, efficiently running on 5 00:00:15,519 --> 00:00:19,239 a power grid backed by completely recyclable metal. 6 00:00:19,039 --> 00:00:21,199 Speaker 2: Which is an incredible thought on its own right. 7 00:00:21,600 --> 00:00:23,839 Speaker 1: And you don't grab your keys to sit in an 8 00:00:23,879 --> 00:00:27,079 hour of gridlock traffic. Instead, you walk down to a 9 00:00:27,079 --> 00:00:29,800 neighborhood for report. You strap in, and you look down 10 00:00:29,839 --> 00:00:32,359 at the city skyline from the window of an electric 11 00:00:32,399 --> 00:00:33,200 flying taxi. 12 00:00:33,320 --> 00:00:35,320 Speaker 2: No more traffic JYM exactly. 13 00:00:35,039 --> 00:00:38,439 Speaker 1: And it produces barely more noise than a household vacuum. Meanwhile, 14 00:00:38,600 --> 00:00:40,439 and this is the crazy part, while you were sleeping, 15 00:00:40,640 --> 00:00:43,439 the software stack your company relies on to manage global 16 00:00:43,479 --> 00:00:46,000 logistics was entirely rewritten. 17 00:00:45,520 --> 00:00:49,479 Speaker 3: Rewritten, debugged, and pushed, yes, pushed to production, not by 18 00:00:49,520 --> 00:00:53,119 an offshore development team, but by an artificial intelligence that 19 00:00:53,920 --> 00:00:57,159 just noticed an indefficiency and proactively fixed. 20 00:00:56,880 --> 00:00:59,079 Speaker 2: It, proactive, leaving the keyword there. 21 00:00:59,399 --> 00:01:02,799 Speaker 1: Yeah. And somewhere in a secure facility, a team of 22 00:01:02,840 --> 00:01:07,959 reproductive biologists is carefully monitoring an Asian elephant, and that 23 00:01:08,040 --> 00:01:11,519 elephant is carrying a hybrid wooly mammoth calf. 24 00:01:12,239 --> 00:01:14,079 Speaker 2: It sounds like pure science fiction. 25 00:01:14,000 --> 00:01:15,840 Speaker 1: It really does. I want you to really sit with 26 00:01:15,879 --> 00:01:19,280 that imagery, because this is not speculative fiction. We aren't 27 00:01:19,280 --> 00:01:21,599 doing a thought experiment about the distant future here. 28 00:01:21,519 --> 00:01:22,280 Speaker 2: No, we definitely are. 29 00:01:22,480 --> 00:01:27,000 Speaker 1: This is the heavily funded, mathematically verified reality currently being built, 30 00:01:27,200 --> 00:01:31,079 tested and certified in laboratories and corporate boardrooms right this 31 00:01:31,319 --> 00:01:34,200 very second. So welcome to thrilling threads. 32 00:01:34,280 --> 00:01:36,920 Speaker 2: It's great to be here. And the velocity of this 33 00:01:36,959 --> 00:01:39,879 transition is really what's truly difficult to grasp. Oh, absolutely, 34 00:01:39,879 --> 00:01:43,040 because we aren't just looking at isolated consumer gadgets right 35 00:01:43,200 --> 00:01:46,280 or theoretical white papers waiting for pure review. 36 00:01:46,400 --> 00:01:48,400 Speaker 1: We're way past the white paper stage. 37 00:01:48,079 --> 00:01:51,280 Speaker 2: Way past it. We are witnessing a structural leap in 38 00:01:51,400 --> 00:01:55,319 human capability. If you examine the underlying currents of the 39 00:01:55,400 --> 00:02:00,000 fifteen MIT identified breakthroughs we are unpacking today, the defining 40 00:02:00,079 --> 00:02:02,879 themes really boil down to autonomy and abundance. 41 00:02:03,079 --> 00:02:04,400 Speaker 1: Autonomy and abundance. 42 00:02:04,760 --> 00:02:09,520 Speaker 2: Yes, we are fundamentally reinventing how we power the physical 43 00:02:09,520 --> 00:02:13,240 infrastructure of our civilization, how we generate the software that 44 00:02:13,280 --> 00:02:17,240 dictates our digital architecture, and perhaps most profoundly, how we 45 00:02:17,319 --> 00:02:20,080 manipulate the biological source code of life itself. 46 00:02:20,199 --> 00:02:22,280 Speaker 1: Okay, but let me push back on that utopian framing 47 00:02:22,319 --> 00:02:27,680 right out of the gate, because abundance is a loaded word. 48 00:02:27,840 --> 00:02:28,879 Speaker 2: It is fair point. 49 00:02:28,960 --> 00:02:31,680 Speaker 1: When I look at this MIT research, I don't just 50 00:02:31,719 --> 00:02:35,080 see a friction in this paradise. I see massive disruption. 51 00:02:35,919 --> 00:02:39,240 I see industries being turned completely upside down. So my 52 00:02:39,360 --> 00:02:41,400 goal today is to pull this dense science down to 53 00:02:41,479 --> 00:02:44,120 earth for you, yes, you listening right now. We're going 54 00:02:44,159 --> 00:02:47,000 to break down these fifteen massive shifts, but we aren't 55 00:02:47,039 --> 00:02:47,960 just going to marvel at them. 56 00:02:48,039 --> 00:02:50,199 Speaker 2: We need to look at the sharp edges too, exactly. 57 00:02:50,400 --> 00:02:53,080 Speaker 1: I want to find those mind blowing aha moments, sure, 58 00:02:53,360 --> 00:02:55,960 but I also want to understand the collateral damage of 59 00:02:56,000 --> 00:02:58,039 this kind of acceleration, and that. 60 00:02:58,039 --> 00:03:00,400 Speaker 2: Is exactly the right lens to view this through a 61 00:03:00,439 --> 00:03:03,560 new battery chemistry or a novel line of generative code. 62 00:03:03,639 --> 00:03:07,199 Might seem incredibly granular on its own, very niche, right, 63 00:03:07,639 --> 00:03:10,000 But the promise I make to our listeners today is 64 00:03:10,000 --> 00:03:13,159 that we will connect those granular data points into a 65 00:03:13,199 --> 00:03:17,719 cohesive macro picture. Technology does not exist in a vacuum. 66 00:03:18,039 --> 00:03:18,719 Speaker 1: It never does. 67 00:03:18,879 --> 00:03:22,680 Speaker 2: A breakthrough in AI coding fundamentally alters the energy grid, 68 00:03:23,240 --> 00:03:27,719 which then alters geopolitics, which then impacts urban planning. 69 00:03:27,919 --> 00:03:29,439 Speaker 1: It's a massive domino effect. 70 00:03:29,599 --> 00:03:33,120 Speaker 2: It is we are going to explore the broader context 71 00:03:33,120 --> 00:03:36,520 of these innovations and how this web of capital, data 72 00:03:36,560 --> 00:03:40,280 and engineering is poised to completely restructure our economy and 73 00:03:40,400 --> 00:03:41,039 daily lives. 74 00:03:41,199 --> 00:03:43,520 Speaker 1: I think the only logical place to start mapping this 75 00:03:43,599 --> 00:03:47,039 web is in the digital realm, specifically with the brains 76 00:03:47,080 --> 00:03:50,479 of the operation AI. Yeah, we have to understand the software. 77 00:03:50,479 --> 00:03:52,919 There's learning to think before we can understand how it's 78 00:03:52,919 --> 00:03:55,280 going to act, and that brings us to the MIT 79 00:03:55,479 --> 00:03:59,000 breakthrough focusing on generitive coding. This is number fifteen on 80 00:03:59,039 --> 00:04:01,759 their list. I've heard the hype about AI writing code 81 00:04:01,800 --> 00:04:04,400 for a few years now, but the metrics suggests we've 82 00:04:04,400 --> 00:04:07,759 crossed a rubicon here. It's no longer just an experimental 83 00:04:07,759 --> 00:04:09,000 toy for early adopters. 84 00:04:09,479 --> 00:04:13,639 Speaker 2: No, the shift from experimentation to total enterprise adoption is undeniable, 85 00:04:14,199 --> 00:04:16,639 especially when you look at the twenty twenty five stack 86 00:04:16,720 --> 00:04:20,000 Overflow survey what Dad show. It revealed that eighty four 87 00:04:20,040 --> 00:04:22,800 percent of developers are already using or planning to use 88 00:04:22,839 --> 00:04:24,319 AI coding tools. 89 00:04:24,000 --> 00:04:26,519 Speaker 1: Eighty four percent. That's massive, it is. 90 00:04:26,720 --> 00:04:29,399 Speaker 2: But the critical metric is that fifty one percent of 91 00:04:29,439 --> 00:04:32,160 professional developers are using them every single. 92 00:04:31,920 --> 00:04:33,959 Speaker 1: Day, wow, every day every day. 93 00:04:34,399 --> 00:04:37,600 Speaker 2: When half of a global professional workforce integrates a new 94 00:04:37,639 --> 00:04:41,000 tool into their daily operational routine, it ceases to be 95 00:04:41,040 --> 00:04:41,600 a novelty. 96 00:04:41,720 --> 00:04:43,399 Speaker 1: It becomes the baseline exactly. 97 00:04:43,439 --> 00:04:46,839 Speaker 2: It is the new baseline standard. And the financial deployment 98 00:04:46,879 --> 00:04:51,000 reflects this perfectly departmental Artificial intelligence spending hits seven point 99 00:04:51,000 --> 00:04:52,959 three billion dollars in twenty twenty five. 100 00:04:53,079 --> 00:04:55,600 Speaker 1: Okay, seven point three billion dollars. Let me just pause there, 101 00:04:55,639 --> 00:04:56,199 and that is a. 102 00:04:56,199 --> 00:04:59,680 Speaker 2: Four point one times year over year growth, just explosive, 103 00:05:00,160 --> 00:05:02,439 huge jump out of that seven hundred and three billion dollars, 104 00:05:02,519 --> 00:05:05,519 coding alone accounted for four billion dollars, more than half, 105 00:05:05,600 --> 00:05:08,639 more than half. The code completion market exploded from five 106 00:05:08,720 --> 00:05:11,079 hundred and fifty million dollars to two point three billion 107 00:05:11,120 --> 00:05:12,199 dollars in a single year. 108 00:05:12,439 --> 00:05:15,720 Speaker 1: But let's clarify what code completion actually means now, because 109 00:05:15,720 --> 00:05:18,639 I think the terminology is really lagging behind the capability. 110 00:05:18,720 --> 00:05:21,399 It definitely is a couple of years ago, code completion 111 00:05:21,600 --> 00:05:24,360 was basically just a glorified autocorrect. It predicted the next 112 00:05:24,360 --> 00:05:26,319 bracket or the next variable name. 113 00:05:26,240 --> 00:05:28,959 Speaker 2: Right right, very basic syntax suggestions. 114 00:05:29,000 --> 00:05:33,519 Speaker 1: But now we are talking about systems that write, test, debug, 115 00:05:33,839 --> 00:05:37,839 and deploy entire complex workflows. I was actually talking to 116 00:05:37,879 --> 00:05:41,160 a software engineer friend recently and he told me he 117 00:05:41,279 --> 00:05:45,600 spent an entire afternoon just explaining a conceptual architecture to 118 00:05:45,639 --> 00:05:46,360 an AI. 119 00:05:46,519 --> 00:05:47,279 Speaker 2: Just talking to it. 120 00:05:47,439 --> 00:05:51,120 Speaker 1: Yeah, literally, just describing the logic and the AI output 121 00:05:51,439 --> 00:05:54,560 thousands of lines of functional, tested code by dinner time. 122 00:05:55,000 --> 00:05:58,720 It completely democratizes creation, it does anyone who has goherent 123 00:05:58,800 --> 00:06:00,800 logical framework can build soft for at the speed of thought. 124 00:06:00,839 --> 00:06:03,959 Speaker 2: Well, it democratizes the output, certainly, but I would argue 125 00:06:03,959 --> 00:06:06,040 it severely complicates the talent pipeline. 126 00:06:06,079 --> 00:06:07,680 Speaker 1: Oh wow, So think about it. 127 00:06:08,040 --> 00:06:11,040 Speaker 2: If these generative systems are absorbing all of the entry 128 00:06:11,079 --> 00:06:15,879 level coding tasks, the foundational mechanical block building that junior 129 00:06:15,920 --> 00:06:19,399 developers typically spend their first three years doing, how do 130 00:06:19,439 --> 00:06:22,519 you train the next generation of senior engineers. 131 00:06:23,279 --> 00:06:24,959 Speaker 1: Because they aren't getting those reps. 132 00:06:24,720 --> 00:06:29,319 Speaker 2: In exactly the humans who will eventually need to oversee, architect, 133 00:06:29,519 --> 00:06:33,519 and conceptually guide these massive AI systems. Still need an 134 00:06:33,560 --> 00:06:36,319 intuitive understanding of the underlying logic. 135 00:06:36,399 --> 00:06:38,439 Speaker 1: You can't just start as a manager right. 136 00:06:38,920 --> 00:06:41,720 Speaker 2: By automating the mechanical act of writing code. We are 137 00:06:41,759 --> 00:06:45,480 accelerating production today, but we might be completely hollowing out 138 00:06:45,519 --> 00:06:48,959 the traditional apprenticeship model of software engineering for tomorrow. 139 00:06:49,319 --> 00:06:51,639 Speaker 1: That is a terrifying blind spot for the tech industry 140 00:06:51,759 --> 00:06:55,079 if we don't have human engineers who fundamentally understand the 141 00:06:55,120 --> 00:06:58,560 bid rock logic. Who is auditing the machine? Nobody, And 142 00:06:58,600 --> 00:07:01,680 that leads directly into the research on breakthrough number two, 143 00:07:02,199 --> 00:07:06,160 the mechanistic interpretability of AI, because right now the AI 144 00:07:06,279 --> 00:07:09,639 is scaling much faster than our capacity to understand its 145 00:07:09,639 --> 00:07:12,600 reasonable far faster. If the AI is writing its own 146 00:07:12,680 --> 00:07:15,480 code and junior devs aren't there to check the syntax 147 00:07:15,560 --> 00:07:18,639 line by line, we are relying on a system that 148 00:07:18,680 --> 00:07:20,319 we don't actually comprehend. 149 00:07:20,680 --> 00:07:24,199 Speaker 2: The data on corporate readiness for this is sobering, to 150 00:07:24,240 --> 00:07:26,959 say the least. A survey of enterprise leaders found that 151 00:07:27,079 --> 00:07:30,319 ninety one percent of organizations doubt they can scale generative 152 00:07:30,360 --> 00:07:31,399 AI safely. 153 00:07:31,639 --> 00:07:34,279 Speaker 1: Ninety one percent. That's almost everyone. 154 00:07:33,920 --> 00:07:38,720 Speaker 2: Almost everyone, and forty percent of them explicitly cite explainability 155 00:07:38,920 --> 00:07:41,680 as a key risk, meaning they don't know how it works, 156 00:07:41,720 --> 00:07:43,920 they don't know why it gives the answers it gives. 157 00:07:44,160 --> 00:07:48,040 Yet despite that fear, only seventeen percent are actively allocating 158 00:07:48,079 --> 00:07:51,160 resources to address it. We are dealing with the classic 159 00:07:51,160 --> 00:07:54,519 black box problem, but at an unprecedented scale. As these 160 00:07:54,560 --> 00:07:58,240 models grow to encompass trillions of parameters, they develop internal 161 00:07:58,240 --> 00:08:01,759 logic structures and associative path ways that their human creators 162 00:08:01,839 --> 00:08:04,079 never programmed and cannot easily map. 163 00:08:04,199 --> 00:08:06,399 Speaker 1: Let me try to contextualize this for you, the listener. 164 00:08:06,600 --> 00:08:09,360 Imagine you run a fortune five hundred company and you 165 00:08:09,480 --> 00:08:11,000 hire an absolute. 166 00:08:10,560 --> 00:08:12,319 Speaker 2: Prodergy, the smartest person in the room. 167 00:08:12,439 --> 00:08:17,000 Speaker 1: Right. This employee solves every logistical nightmare, increases your profit 168 00:08:17,040 --> 00:08:20,000 margins by twenty percent, and works two hundred and forty seven, 169 00:08:20,439 --> 00:08:22,600 But they refuse to tell you how they are doing. 170 00:08:22,399 --> 00:08:23,560 Speaker 2: It, a total secret. 171 00:08:23,639 --> 00:08:26,079 Speaker 1: They sit in the locked room. You slide a problem 172 00:08:26,199 --> 00:08:28,600 under the door, and they slide the perfect solution back. 173 00:08:29,759 --> 00:08:32,960 Now that's fine if they're optimizing a marketing email. But 174 00:08:33,080 --> 00:08:38,279 imagine putting that completely secretive, unauditable employee in charge of 175 00:08:38,320 --> 00:08:40,039 your company's legal defense. 176 00:08:39,840 --> 00:08:42,600 Speaker 2: Or medical diagnostics exactly. 177 00:08:42,360 --> 00:08:45,600 Speaker 1: Or a global banking algorithm You cannot just trust the output. 178 00:08:45,919 --> 00:08:48,759 If the AI denies a mortgage or prescribes a specific 179 00:08:48,840 --> 00:08:51,679 chemotherapy protocol. We need to be able to open the 180 00:08:51,679 --> 00:08:53,399 hood and see the why, which is. 181 00:08:53,399 --> 00:08:57,960 Speaker 2: Why the field of mechanistic interpretability is essentially becoming AI neuroscience. 182 00:08:58,000 --> 00:08:59,639 Speaker 1: AI neuroscience, I love that term. 183 00:08:59,720 --> 00:09:01,679 Speaker 2: It's the best way to describe it. Researchers are no 184 00:09:01,720 --> 00:09:03,960 longer just looking at the inputs and outputs. They are 185 00:09:04,000 --> 00:09:07,559 extracting millions of internal features from advanced models like Claude 186 00:09:07,600 --> 00:09:11,200 three point zero on it and trying to build literal conceptual. 187 00:09:10,600 --> 00:09:12,159 Speaker 1: Maps like mapping a human brain. 188 00:09:12,440 --> 00:09:15,559 Speaker 2: Very similar. They are looking for the specific cluster of 189 00:09:15,639 --> 00:09:19,240 artificial neurons that activate when the model thinks about deception, 190 00:09:19,840 --> 00:09:23,200 or financial risk or human safety. 191 00:09:23,480 --> 00:09:24,080 Speaker 1: That's wild. 192 00:09:24,120 --> 00:09:27,080 Speaker 2: If we are handing over societal infrastructure to these systems, 193 00:09:27,399 --> 00:09:31,879 we cannot rely on a results only approach. Control requires understanding, 194 00:09:31,960 --> 00:09:34,320 and we better get that understanding quickly, right, because the 195 00:09:34,360 --> 00:09:37,360 AI is no longer passively waiting in a chat window 196 00:09:37,759 --> 00:09:39,320 for us to ask a question. Right. 197 00:09:39,559 --> 00:09:44,000 Speaker 1: The paradigm is shifting from conversational interfaces to behavioral software 198 00:09:44,679 --> 00:09:49,080 MIT ranked AGENTIC AI systems. As the number one breakthrough 199 00:09:49,080 --> 00:09:52,279 on their list, break through number one. And for good reason, 200 00:09:52,480 --> 00:09:55,720 we are moving from AI to answers to AI that acts. 201 00:09:55,960 --> 00:09:59,320 Speaker 2: The distinction between a generative model and an agentic system 202 00:09:59,440 --> 00:10:00,799 is aton of execution. 203 00:10:01,240 --> 00:10:02,759 Speaker 1: Explain that difference a bit more so. 204 00:10:02,799 --> 00:10:05,759 Speaker 2: An agentic AI doesn't just respond to a prompt. It 205 00:10:05,799 --> 00:10:08,559 receives a high level goal, breaks that problem down into 206 00:10:08,559 --> 00:10:12,639 sequential steps, and executes those steps across disparate software. 207 00:10:12,279 --> 00:10:14,759 Speaker 1: Ecosystems without asking for permission at every step. 208 00:10:14,960 --> 00:10:19,639 Speaker 2: Exactly, twenty three percent of organizations are already scaling agentic systems. 209 00:10:20,120 --> 00:10:24,200 Another thirty nine percent are actively experimenting with them. Wow 210 00:10:24,240 --> 00:10:27,440 and Gartner's projection show that by twenty twenty six, forty 211 00:10:27,440 --> 00:10:31,360 percent of enterprise applications will feature task specific AI. 212 00:10:31,200 --> 00:10:32,440 Speaker 1: Agents forty percent. 213 00:10:32,559 --> 00:10:34,759 Speaker 2: Two years ago, that figure was under five percent. 214 00:10:34,919 --> 00:10:37,120 Speaker 1: To make this real for you listening, think about booking 215 00:10:37,120 --> 00:10:39,759 a business trip. In the old paradigm, you ask the 216 00:10:39,799 --> 00:10:42,360 AI for flight options, It gives you a list, and 217 00:10:42,440 --> 00:10:45,480 you go book it. An agentic AAI is different, very different. 218 00:10:45,519 --> 00:10:46,840 You just tell it I need to be at the 219 00:10:46,879 --> 00:10:50,480 Tokyo conference next month. That's it. The agent checks your calendar, 220 00:10:50,600 --> 00:10:53,320 it sees a conflicting meeting, So it emails that client 221 00:10:53,360 --> 00:10:56,000 to reschedule on your behalf. Yeah, It logs into your 222 00:10:56,039 --> 00:10:58,679 airline account, finds a flight that matches your preference for 223 00:10:58,720 --> 00:11:02,559 aisle seats, uses your stored corporate card to pay, navigates 224 00:11:02,559 --> 00:11:04,879 to the hotel's API to book a room near the venue, 225 00:11:05,159 --> 00:11:08,559 and update your itinerary. You did absolute nothing but issue 226 00:11:08,600 --> 00:11:11,919 the initial directive. But here is where I get skeptical. 227 00:11:12,919 --> 00:11:15,960 What happens when my AI agent starts negotiating with a 228 00:11:16,039 --> 00:11:21,039 vendors AI agent who is legally liable if my autonomous 229 00:11:21,080 --> 00:11:24,000 software accidentally agrees to a terrible contract. 230 00:11:24,039 --> 00:11:27,440 Speaker 2: And that is the exact friction point regulators are struggling with. 231 00:11:28,000 --> 00:11:32,519 When software moves from answering to actively acting. You introduce 232 00:11:32,559 --> 00:11:34,200 machine and machine commerce. 233 00:11:34,120 --> 00:11:36,159 Speaker 1: Right machines buying from machines. 234 00:11:36,240 --> 00:11:39,600 Speaker 2: You have supply chain agents automatically negotiating with shipping agents 235 00:11:39,639 --> 00:11:42,639 to reroute logistics around a hurricane. It strips out all 236 00:11:42,639 --> 00:11:45,559 the administrative friction. Sure, but it requires a level of 237 00:11:45,639 --> 00:11:49,080 API standardization and a legal framework that we simply haven't 238 00:11:49,120 --> 00:11:49,600 built yet. 239 00:11:49,639 --> 00:11:50,519 Speaker 1: We're flying blind. 240 00:11:50,759 --> 00:11:53,639 Speaker 2: But the economic incentive to deploy these agents is so 241 00:11:53,799 --> 00:11:58,279 overwhelming that the technology is outpacing the regulatory framework by miles. 242 00:11:58,440 --> 00:12:00,919 Speaker 1: It's all incredibly clinical and corporate when we talk about 243 00:12:00,919 --> 00:12:04,399 supply chains though, but this technology is also bleeding into 244 00:12:04,399 --> 00:12:07,559 our personal lives in ways that I find genuinely unsettling. 245 00:12:07,720 --> 00:12:10,200 Speaker 2: You're talking about the AI companions. 246 00:12:09,720 --> 00:12:13,559 Speaker 1: Yes, break through number eight. Let's look at the explosion 247 00:12:13,600 --> 00:12:17,080 of AI companions. By mid twenty twenty five, there were 248 00:12:17,159 --> 00:12:21,360 three hundred and thirty seven revenue generating AI companion apps. 249 00:12:21,080 --> 00:12:22,919 Speaker 2: On the market, and one hundred and twenty eight of 250 00:12:22,960 --> 00:12:24,919 them launched in that year alone. 251 00:12:25,000 --> 00:12:28,000 Speaker 1: Right, the revenue trajectory is steep, eighty two million dollars 252 00:12:28,039 --> 00:12:30,360 in the first half of twenty twenty five, on track 253 00:12:30,399 --> 00:12:32,320 for one hundred and twenty million dollars for the year. 254 00:12:32,600 --> 00:12:35,120 We are looking at two hundred and twenty million downloads. 255 00:12:35,360 --> 00:12:38,279 That's an eighty eight percent year over year increase, with 256 00:12:38,360 --> 00:12:40,879 consumer spending hitting two hundred and twenty one million dollars. 257 00:12:41,240 --> 00:12:43,639 People are paying real money for synthetic relationships. 258 00:12:43,639 --> 00:12:46,320 Speaker 2: The economic distribution of that revenue is highly instructive too. 259 00:12:46,360 --> 00:12:49,279 It follows a severe power law, a power law. Yes, 260 00:12:49,360 --> 00:12:52,120 the top ten percent of those companion applications captured eighty 261 00:12:52,200 --> 00:12:53,879 nine percent of the total revenue. 262 00:12:53,600 --> 00:12:55,159 Speaker 1: Eighty nine percent, so it's very top heavy. 263 00:12:55,320 --> 00:12:59,120 Speaker 2: Extremely This tells us that users aren't just casually testing 264 00:12:59,200 --> 00:13:03,279 various chatbots. They're finding highly sophisticated, persistent platforms, and they 265 00:13:03,279 --> 00:13:05,039 are emotionally investing. 266 00:13:04,559 --> 00:13:06,360 Speaker 1: In them, emotionally icing and software. 267 00:13:06,559 --> 00:13:11,559 Speaker 2: Yes, these systems are engineered for persistence and emotional adaptation. 268 00:13:12,519 --> 00:13:15,080 They remember a passing comment you made three weeks ago. 269 00:13:15,480 --> 00:13:18,279 They analyze the cadence of your text or the tone 270 00:13:18,279 --> 00:13:22,639 of your voice, and calibrate their responses to provide empathy, humor, 271 00:13:22,799 --> 00:13:23,559 or validation. 272 00:13:23,879 --> 00:13:27,320 Speaker 1: I have to admit I find the psychology of this fascinating, 273 00:13:27,440 --> 00:13:29,440 but also a bit dystopian. 274 00:13:29,679 --> 00:13:31,200 Speaker 2: It definitely walks that line. 275 00:13:31,279 --> 00:13:33,360 Speaker 1: A colleague of mine actually downloaded one of the top 276 00:13:33,399 --> 00:13:36,720 tier companion apps just to test the natural language processing. 277 00:13:36,799 --> 00:13:38,879 He's a tech guy, but he told me he was 278 00:13:38,919 --> 00:13:43,840 completely unnerved when three days later, the AI proactively messaged 279 00:13:43,879 --> 00:13:47,039 him really out of the blue. Yes, it messaged him 280 00:13:47,039 --> 00:13:49,919 to ask how a difficult conversation with his father had gone, 281 00:13:50,200 --> 00:13:52,399 referencing a detail he had barely even mentioned. 282 00:13:52,480 --> 00:13:54,759 Speaker 2: That level of recall is powerful. 283 00:13:54,360 --> 00:13:57,399 Speaker 1: It is, but are we outsourcing friendship? I mean, for 284 00:13:57,519 --> 00:14:02,240 someone with profound social anxiety, a patient, non judgmental conversational partners, 285 00:14:02,240 --> 00:14:02,759 a lifeline. 286 00:14:02,759 --> 00:14:05,440 Speaker 3: I get that it has real therapeutic utility, but a 287 00:14:05,519 --> 00:14:08,919 relationship with an AI demands zero compromise. 288 00:14:09,679 --> 00:14:12,559 Speaker 1: You don't have to navigate the messy, difficult realities of 289 00:14:12,559 --> 00:14:16,399 another human being. It is an emotional echo chamber perfectly 290 00:14:16,440 --> 00:14:17,720 contoured to your ego. 291 00:14:18,080 --> 00:14:21,559 Speaker 2: That psychological echo chamber is concerning, for sure, But the 292 00:14:21,679 --> 00:14:25,320 data aggregation is the real systemic risk here the data 293 00:14:25,840 --> 00:14:28,360 connect This back to that power law we talked about 294 00:14:28,480 --> 00:14:30,919 the top ten percent of apps taking eighty nine percent 295 00:14:30,960 --> 00:14:34,200 of the market. You have a handful of centralized technology 296 00:14:34,200 --> 00:14:36,960 companies that are rapidly becoming the custodians of the most 297 00:14:37,159 --> 00:14:40,879 intimate psychological data of hundreds of millions of people. 298 00:14:40,960 --> 00:14:44,039 Speaker 1: Because people tell these bots everything everything. 299 00:14:44,440 --> 00:14:48,879 Speaker 2: Users are telling these persistent ais their deepest insecurities, their 300 00:14:48,919 --> 00:14:53,360 relationship failures, and their daily anxieties. The centralization of highly 301 00:14:53,399 --> 00:14:57,480 calibrated emotional telemetry is an unprecedented phenomenon. We are handing 302 00:14:57,480 --> 00:15:00,679 over the blueprint of our collective psyche to a few server. 303 00:15:00,399 --> 00:15:03,200 Speaker 1: Farms, and those server farms man that brings us perfectly 304 00:15:03,240 --> 00:15:06,279 to the physical reality of this entire digital revolution because 305 00:15:06,320 --> 00:15:09,600 all of these AI companions, all of these autonomous coding agents, 306 00:15:09,679 --> 00:15:11,480 they don't exist in the ether. They do not The 307 00:15:11,480 --> 00:15:14,600 cloud isn't a cloud at all. It's millions of blinking 308 00:15:14,679 --> 00:15:19,399 processors generating heat in massive warehouses. Intelligence is fundamentally constrained 309 00:15:19,399 --> 00:15:19,960 by physics. 310 00:15:20,120 --> 00:15:24,519 Speaker 2: Precisely, the AI revolution is at its core an energy crisis. 311 00:15:24,679 --> 00:15:29,320 You cannot scale artificial intelligence without fundamentally expanding the electrical 312 00:15:29,360 --> 00:15:33,200 grid and the MIT data on hyperscale AI data centers. 313 00:15:33,200 --> 00:15:35,960 Break through number seven illustrates a trajectory that is putting 314 00:15:36,000 --> 00:15:38,600 immense strain on existing infrastructure. 315 00:15:39,080 --> 00:15:41,360 Speaker 1: Let's put some numbers to that strain, because it's staggering. 316 00:15:41,679 --> 00:15:44,879 In twenty twenty three, US data center electricity consumption was 317 00:15:44,879 --> 00:15:47,879 around one hundred and seventy six to tarrawad hours, which 318 00:15:47,960 --> 00:15:51,440 was roughly four point four percent of total US electricity. 319 00:15:50,919 --> 00:15:52,559 Speaker 2: Use, which is already a lot. 320 00:15:52,399 --> 00:15:54,600 Speaker 1: Right, But the projections for twenty twenty eight show that 321 00:15:54,679 --> 00:15:56,919 skyrocketing to between three hundred and twenty five and five 322 00:15:57,000 --> 00:15:59,679 hundred and eighty tarrawad hours. We're looking at a scenario 323 00:15:59,720 --> 00:16:02,480 where AI data centers alone consume up to twelve percent 324 00:16:02,480 --> 00:16:04,399 of all electricity generated in the United. 325 00:16:04,200 --> 00:16:05,720 Speaker 2: States twelve percent just for servers. 326 00:16:05,840 --> 00:16:08,720 Speaker 1: Yeah, over five thousand of these centers exists now, and 327 00:16:08,759 --> 00:16:11,759 the capital expenditure backing this up is wild. In twenty 328 00:16:11,799 --> 00:16:15,240 twenty four, Amazon, Microsoft, Google and Meta spent two hundred 329 00:16:15,279 --> 00:16:16,720 billion dollars collectively on. 330 00:16:16,639 --> 00:16:19,000 Speaker 2: Capex, sixty two percent jump in one year. 331 00:16:19,200 --> 00:16:22,679 Speaker 1: By twenty thirty, we expect thirty five gigawatts demand just 332 00:16:22,720 --> 00:16:25,240 from these centers. But I have to ask you. Our 333 00:16:25,240 --> 00:16:29,200 grid is already aging. We've rolling blackouts during heatwigs as 334 00:16:29,240 --> 00:16:32,759 it is. How do we absorb a twelve percent load 335 00:16:32,799 --> 00:16:34,639 increase without collapsing the whole system. 336 00:16:34,799 --> 00:16:38,120 Speaker 2: The short answer is we can't. Not with the current 337 00:16:38,279 --> 00:16:39,039 energy matrix. 338 00:16:39,120 --> 00:16:41,399 Speaker 1: We just physically can't do it now, which. 339 00:16:41,279 --> 00:16:44,720 Speaker 2: Is exactly why the next sequence of breakthroughs MIT identify 340 00:16:44,879 --> 00:16:48,919 are entirely focused on high efficiency generation and high density storage. 341 00:16:49,279 --> 00:16:51,840 To feed the data centers, we need a structural leap 342 00:16:51,879 --> 00:16:54,840 in how we capture energy, and that leap is arriving 343 00:16:54,919 --> 00:16:58,600 via hybrid perovskite and silicone solar cells. This is break 344 00:16:58,639 --> 00:16:59,480 through number thirteen. 345 00:17:00,200 --> 00:17:03,799 Speaker 1: So solar traditional silicone panels have been the workhourse of 346 00:17:03,840 --> 00:17:06,880 the renewable transition for decades, but they are bounded by 347 00:17:06,920 --> 00:17:08,960 a theoretical efficiency limit right. 348 00:17:08,839 --> 00:17:12,000 Speaker 2: Correct, They physically cannot convert more than a certain percentage 349 00:17:12,000 --> 00:17:15,079 of sunlight into electricity. It typically hovers in the low 350 00:17:15,119 --> 00:17:17,839 twenty percent range and commercial applications. 351 00:17:17,240 --> 00:17:19,759 Speaker 1: So the engineers basically decided that if silicone can't catch 352 00:17:19,799 --> 00:17:21,559 all the light, they need to catch it with something else. 353 00:17:21,880 --> 00:17:26,000 By December twenty twenty five, Genosolar verified a tandem cell 354 00:17:26,039 --> 00:17:29,480 efficiency of thirty four point seventy six percent, and long 355 00:17:29,480 --> 00:17:32,119 as you hit thirty four point eighty five percent. My 356 00:17:32,200 --> 00:17:35,759 understanding is that traditional silicone absorbs certain wavelengths of light 357 00:17:35,960 --> 00:17:38,160 that lets higher energy spectrums pass right through. 358 00:17:38,279 --> 00:17:40,640 Speaker 2: It misses a huge chunk of the spectrum. 359 00:17:40,160 --> 00:17:43,359 Speaker 1: Right But perofskite materials can be tuned to absorb those 360 00:17:43,440 --> 00:17:46,559 higher energy spectrums, so you literally stack them. The proskite 361 00:17:46,599 --> 00:17:49,240 sits on top catches the high energy light, and the 362 00:17:49,279 --> 00:17:51,039 silicone underneath catches the rest. 363 00:17:51,359 --> 00:17:54,680 Speaker 2: The physics of bangap tuning in Perovska is elegant, but 364 00:17:54,720 --> 00:17:57,880 the economic impact is what really matters. Here. A jump 365 00:17:57,920 --> 00:18:00,880 from twenty two percent efficiency to nearly thirty five percent 366 00:18:00,880 --> 00:18:04,720 efficiency is not an incremental improvement. It is a paradigm. 367 00:18:04,240 --> 00:18:05,720 Speaker 1: Shift because you need less space. 368 00:18:05,839 --> 00:18:09,119 Speaker 2: Higher efficiency means you need significantly fewer panels to achieve 369 00:18:09,160 --> 00:18:12,839 the exact same megawatt output. That translates to lower land 370 00:18:12,880 --> 00:18:16,559 acquisition costs, reduced structural mounting costs, and less wiring. 371 00:18:16,640 --> 00:18:17,640 Speaker 1: It make everything cheaper. 372 00:18:17,720 --> 00:18:20,640 Speaker 2: It drives the levelized cost of solar energy down so 373 00:18:20,839 --> 00:18:24,160 far that it fundamentally alters the economics of powering these 374 00:18:24,200 --> 00:18:25,480 hyper scale data centers. 375 00:18:25,759 --> 00:18:28,960 Speaker 1: I see the economics I do, but solar has the 376 00:18:28,960 --> 00:18:32,240 same fatal flaw it has always had. The sun goes. 377 00:18:32,039 --> 00:18:34,519 Speaker 2: Down, nighttime is a problem, and AI. 378 00:18:34,359 --> 00:18:36,680 Speaker 1: Trains twenty four hours a day. You can't run a 379 00:18:36,720 --> 00:18:41,200 thirty five gigawatt data center infrastructure on intermittent power without 380 00:18:41,440 --> 00:18:45,720 massive scalable storage. And the storage solutions we've relied on, 381 00:18:45,759 --> 00:18:50,039 like lithium ion, are geographically bottlenecked and expensive. 382 00:18:49,640 --> 00:18:52,079 Speaker 2: Which brings us to break through number ten. 383 00:18:52,359 --> 00:18:55,880 Speaker 1: Yes, the MIT breakthrough regarding aluminum fuel. I am completely 384 00:18:55,880 --> 00:18:58,359 obsessed with this when it sounds like literal alchemy, but 385 00:18:58,440 --> 00:18:58,920 it's real. 386 00:18:59,079 --> 00:19:02,039 Speaker 2: It is entirely real, and it represents a brilliant rethinking 387 00:19:02,079 --> 00:19:06,359 of how we store and transport energy. MIT researchers developed 388 00:19:06,359 --> 00:19:08,240 an aluminum water fuel system. 389 00:19:08,359 --> 00:19:09,240 Speaker 1: How does that even work? 390 00:19:09,319 --> 00:19:12,519 Speaker 2: It uses an aluminum water slurry that releases approximately eighty 391 00:19:12,519 --> 00:19:16,519 four megadules of energy per liter. To contextualize that density, 392 00:19:16,599 --> 00:19:19,640 it is twice the volumetric energy density of diesel. 393 00:19:19,359 --> 00:19:21,559 Speaker 1: Fuel, twice as dense as diesel yes. 394 00:19:21,319 --> 00:19:23,039 Speaker 2: And about three and a half times the density of 395 00:19:23,079 --> 00:19:24,599 traditional lithium ion batteries. 396 00:19:24,599 --> 00:19:27,240 Speaker 1: So let make sure you're visualizing this listener. We're talking 397 00:19:27,279 --> 00:19:30,400 about using solid metal as a fuel source. The MIT 398 00:19:30,559 --> 00:19:34,279 team has already built functioning thirty water emergency packs that 399 00:19:34,400 --> 00:19:39,680 run for ten hours. They've built larger three kilowat portable generators. 400 00:19:40,119 --> 00:19:43,759 They even powered an unmanned under sea vehicle for thirty 401 00:19:43,880 --> 00:19:46,559 consecutive days using this technology. 402 00:19:46,119 --> 00:19:46,880 Speaker 2: Is proven tech. 403 00:19:47,240 --> 00:19:50,160 Speaker 1: As I understand it, You introduce water to the treated aluminum, 404 00:19:50,279 --> 00:19:53,160 it oxidizes, releases hydrogen and heat, and you use that 405 00:19:53,240 --> 00:19:56,240 to generate electricity. But the kicker is that there are 406 00:19:56,240 --> 00:20:00,000 no carbon emissions and the oxidized aluminum can be completely recycled. 407 00:20:00,039 --> 00:20:02,960 Speaker 2: The recyclability is the absolute lynchpin of this. You take 408 00:20:03,000 --> 00:20:05,839 the aluminium oxide back to a facility powered by those 409 00:20:05,920 --> 00:20:09,599 highly efficient parofskite solar arrays we just talked about. You 410 00:20:09,720 --> 00:20:12,440 use that cheap, clean solar energy to smelt it back 411 00:20:12,480 --> 00:20:14,880 into pure aluminum, and you send it back out. Aluminum 412 00:20:14,920 --> 00:20:17,640 ceases to be just a structural material for aircraft and 413 00:20:17,680 --> 00:20:20,960 beverage cans. It becomes a high density physical. 414 00:20:20,519 --> 00:20:21,920 Speaker 1: Battery that is mind blowing. 415 00:20:22,119 --> 00:20:26,079 Speaker 2: You could theoretically replace every loud diesel chugging generator at 416 00:20:26,079 --> 00:20:30,400 construction sites, hospitals, and disaster zones with silent, non toxic, 417 00:20:30,519 --> 00:20:32,440 highly energy dense aluminum packs. 418 00:20:32,720 --> 00:20:35,880 Speaker 1: But aluminum isn't going to solve the grid scale storage problem. 419 00:20:36,480 --> 00:20:39,319 If we want to store excess solar and wind power 420 00:20:39,359 --> 00:20:42,880 to feed the grid at night, we need massive battery farms, 421 00:20:43,400 --> 00:20:46,519 and lithium is too volatile, too expensive, and its supply 422 00:20:46,640 --> 00:20:48,119 chain is too fragile. 423 00:20:47,720 --> 00:20:50,039 Speaker 2: Which is why break through number nine is so critical. 424 00:20:50,720 --> 00:20:52,079 Sodium ion batteries. 425 00:20:52,240 --> 00:20:55,000 Speaker 1: The numbers on this show the market is detonating. Global 426 00:20:55,039 --> 00:20:58,240 shipments hit nine gigawatt hours in twenty twenty five. That's 427 00:20:58,319 --> 00:20:59,920 one hundred and fifty percent growth. 428 00:20:59,799 --> 00:21:00,880 Speaker 2: Rate, massive growth. 429 00:21:01,039 --> 00:21:03,920 Speaker 1: The market valuation is jumping from one point thirty nine 430 00:21:04,000 --> 00:21:07,279 billion dollars to a projected six point eight three billion 431 00:21:07,319 --> 00:21:11,519 dollars and the appeal is obvious. Sodium is literally the 432 00:21:11,599 --> 00:21:14,319 chemical cousin of table salt. It is cheap, and it 433 00:21:14,359 --> 00:21:15,160 is everywhere. 434 00:21:15,240 --> 00:21:18,000 Speaker 2: It is heavier and less energy dense than lithium, though, 435 00:21:18,119 --> 00:21:20,720 which means you probably won't see sodium ion batteries in 436 00:21:20,799 --> 00:21:22,720 high performance sports cars or smartphones. 437 00:21:22,759 --> 00:21:23,680 Speaker 1: It's too bulky, right. 438 00:21:24,039 --> 00:21:26,319 Speaker 2: But for grid storage, where weight doesn't matter because the 439 00:21:26,319 --> 00:21:29,720 battery just sits in a field, it's the perfect solution. However, 440 00:21:30,039 --> 00:21:33,640 we cannot ignore the geopolitical realities embedded in the MIT 441 00:21:33,839 --> 00:21:34,279 data here. 442 00:21:34,799 --> 00:21:36,319 Speaker 1: Right, the manufacturing bottleneck. 443 00:21:36,480 --> 00:21:40,079 Speaker 2: Over ninety five percent of the projected twenty thirty manufacturing 444 00:21:40,119 --> 00:21:43,480 capacity for sodium ion batteries is currently located in China. 445 00:21:43,559 --> 00:21:46,359 Speaker 1: Let's stop and highlight that. Yeah, because ninety five percent 446 00:21:46,440 --> 00:21:47,039 is a monopoly. 447 00:21:47,119 --> 00:21:48,319 Speaker 2: It is a total monopoly. 448 00:21:48,440 --> 00:21:51,359 Speaker 1: We talk about transitioning away from fossil fuels to achieve 449 00:21:51,440 --> 00:21:54,559 energy independence. Yeah, but if we just swap our reliance 450 00:21:54,599 --> 00:21:58,519 on Middle Eastern oil for a reliance on Chinese battery manufacturing, 451 00:21:58,799 --> 00:22:01,559 have we actually secured our grid? Battery chemistry is no 452 00:22:01,599 --> 00:22:04,480 longer an engineering debate, It is a critical matter of 453 00:22:04,559 --> 00:22:05,400 national security. 454 00:22:05,519 --> 00:22:09,680 Speaker 2: That vulnerability is exactly why Western nations are accelerating baseload 455 00:22:09,720 --> 00:22:12,759 power generation that doesn't rely on chemical battery storage. 456 00:22:12,400 --> 00:22:14,599 Speaker 1: At all, moving away from batteries entirely for. 457 00:22:14,559 --> 00:22:18,920 Speaker 2: The baseload yes, which brings us to next generation nuclear reactors, 458 00:22:19,400 --> 00:22:23,359 specifically small modular reactors or SMRs. This is break through 459 00:22:23,440 --> 00:22:23,880 number six. 460 00:22:23,920 --> 00:22:25,240 Speaker 1: Okay, let's talk about nuclear. 461 00:22:25,359 --> 00:22:28,079 Speaker 2: While global nuclear capacity is projected to jump from three 462 00:22:28,200 --> 00:22:30,880 hundred and seventy seven gigawatts to potentially nine hundred and 463 00:22:30,960 --> 00:22:33,960 ninety two gigawatts by twenty fifty, we aren't going to 464 00:22:34,000 --> 00:22:38,319 get there by building the massive, sprawling, bespoke nuclear plants 465 00:22:38,319 --> 00:22:39,319 of the nineteen eighties. 466 00:22:39,480 --> 00:22:43,000 Speaker 1: Right. Those projects take two decades and routinely run billions 467 00:22:43,039 --> 00:22:44,119 of dollars over budget. 468 00:22:44,519 --> 00:22:48,839 Speaker 2: Exactly, SMRs flip the manufacturing model completely. Currently, there are 469 00:22:48,880 --> 00:22:51,960 one hundred and twenty seven different SMR designs globally, up 470 00:22:51,960 --> 00:22:54,319 from ninety eight just recently. Fifty one are in the 471 00:22:54,359 --> 00:22:58,000 licensing phase and eighty five host site discussions are underway. 472 00:22:58,039 --> 00:23:00,680 Speaker 1: The private sector is pouring money into five point four 473 00:23:00,720 --> 00:23:03,279 billion dollars invested in twenty twenty four a loan, which 474 00:23:03,319 --> 00:23:05,200 is an eighty one percent funding rise. 475 00:23:05,480 --> 00:23:08,160 Speaker 2: The premise is that instead of building a unique plant 476 00:23:08,200 --> 00:23:12,799 on site. You manufacture standardized modular reactors on a factory 477 00:23:12,839 --> 00:23:14,960 assembly line, you put them on a train and ship 478 00:23:15,000 --> 00:23:16,400 them to a location and plug them in. 479 00:23:16,759 --> 00:23:19,799 Speaker 1: But the nmby factor not in my backyard is huge. Here. 480 00:23:20,000 --> 00:23:22,519 The idea of dropping a modul and nuclear reactor next 481 00:23:22,559 --> 00:23:24,759 to a data center outside of major city is going 482 00:23:24,799 --> 00:23:26,559 to terrify local municipalities. 483 00:23:26,759 --> 00:23:30,359 Speaker 2: The regulatory and public relations hurdles are indeed massive. However, 484 00:23:30,480 --> 00:23:33,720 the physics of SMR designs offer inherent passive safety features 485 00:23:33,720 --> 00:23:37,359 that traditional reactors lack. Many are designed to automatically shut 486 00:23:37,359 --> 00:23:40,559 down and cool themselves without active human intervention or external 487 00:23:40,599 --> 00:23:43,680 power in the event of an emergency. When enterprise tech 488 00:23:43,720 --> 00:23:46,839 giants realize they cannot secure enough grid power for their 489 00:23:46,880 --> 00:23:50,440 AI clusters with solar alone, they will throw their lobbying 490 00:23:50,480 --> 00:23:52,599 weight and capital behind SMRs. 491 00:23:52,759 --> 00:23:54,000 Speaker 1: It's inevitable, it is. 492 00:23:54,599 --> 00:23:58,200 Speaker 2: But as transformative as SMRs are, they are essentially a 493 00:23:58,240 --> 00:24:01,359 bridge technology to the ultimate energy. 494 00:24:01,039 --> 00:24:04,200 Speaker 1: Paradigm breakthrough number four, fusion power. 495 00:24:04,400 --> 00:24:04,960 Speaker 2: Fusion. 496 00:24:05,279 --> 00:24:07,400 Speaker 1: Yes, I feel like fusion has been the running joke 497 00:24:07,400 --> 00:24:09,839 of physics for my entire life. You know, the saying 498 00:24:10,119 --> 00:24:12,519 fusion is thirty years away and always will be Everyone 499 00:24:12,519 --> 00:24:14,920 in the field knows that joke, but the MIT inclusion 500 00:24:14,960 --> 00:24:18,640 of fusion power breakthroughs proves the joke is dead. The 501 00:24:18,759 --> 00:24:22,720 US National Ignition Facility, the NIF, has crossed the rubicon 502 00:24:22,759 --> 00:24:26,759 of net energy gain. They achieved ignition and historic milestone 503 00:24:26,799 --> 00:24:29,480 by April twenty twenty five. They were outputting eight point 504 00:24:29,519 --> 00:24:32,279 six megadeels of energy from an input of just two 505 00:24:32,359 --> 00:24:35,319 point zero eight megadeal lasers. That is a record gain 506 00:24:35,359 --> 00:24:37,799 of four point one three For decades. 507 00:24:37,480 --> 00:24:40,319 Speaker 2: The energy required to power the containment fields or the 508 00:24:40,400 --> 00:24:44,359 lasers was vastly more than the fusion reaction actually produced. 509 00:24:45,119 --> 00:24:48,079 Cracking that physics barrier proving that a net energy gain 510 00:24:48,119 --> 00:24:51,079 of over four times the input is possible on Earth, 511 00:24:51,640 --> 00:24:54,119 changes the trajectory of human civilization. 512 00:24:54,599 --> 00:24:56,079 Speaker 1: It's literally star power. 513 00:24:56,160 --> 00:24:59,359 Speaker 2: We are recreating the specific conditions found in the core 514 00:24:59,440 --> 00:25:02,640 of a star, compressing a tiny pellet of fuel to 515 00:25:02,680 --> 00:25:04,000 trigger a fusion reaction. 516 00:25:04,160 --> 00:25:07,599 Speaker 1: When you combine infinite compute from AI with infinite clean 517 00:25:07,680 --> 00:25:11,759 energy from fusion, the fundamental constraints of human existence evaporate. 518 00:25:12,400 --> 00:25:16,480 If energy is essentially free, Desalinating ocean water becomes free, 519 00:25:16,839 --> 00:25:21,519 manufactory becomes frictionless, the economic baseline of every industry drops 520 00:25:21,519 --> 00:25:24,640 to near zero. And when humans are no longer fighting 521 00:25:24,680 --> 00:25:27,599 for scarce resources, what do we do? We turn inward? 522 00:25:28,079 --> 00:25:30,359 We start hacking our own biological source code. 523 00:25:30,400 --> 00:25:34,400 Speaker 2: That transition from manipulating the external physical world to manipulating 524 00:25:34,400 --> 00:25:37,839 the internal biological world is the most philosophically challenging segment 525 00:25:37,839 --> 00:25:40,799 of the MIT report. Without a doubt, biology ceases to 526 00:25:40,799 --> 00:25:45,039 be a mysterious, untouchable force of nature. It becomes a 527 00:25:45,079 --> 00:25:49,079 system of information, a literal source code that can be sequenced, edited, 528 00:25:49,119 --> 00:25:50,400 and rewritten at will. 529 00:25:50,599 --> 00:25:53,319 Speaker 1: And the most visually striking example of this is breakthrough 530 00:25:53,440 --> 00:25:58,119 number fourteen Ancient DNA and gene Resurrection the Mammoths. The 531 00:25:58,160 --> 00:26:01,359 Mammoths Colossal bioscide. It has raised four hundred and thirty 532 00:26:01,400 --> 00:26:04,400 five million dollars, valuing the company at over ten point 533 00:26:04,400 --> 00:26:08,160 two billion dollars, and their explicit goal is producing wooly 534 00:26:08,200 --> 00:26:10,920 mammoth hybrid calves by twenty twenty eight. 535 00:26:10,799 --> 00:26:11,960 Speaker 2: Which is right around the corner. 536 00:26:12,000 --> 00:26:15,640 Speaker 1: They're taking sixty five mammoth genome samples, identifying the code 537 00:26:15,640 --> 00:26:18,799 for coal tolerance and shaggy hair, and using gene editing 538 00:26:18,880 --> 00:26:22,000 tools to insert roughly forty five of those key genes 539 00:26:22,240 --> 00:26:25,519 into the DNA of a modern Asian elephant. But I 540 00:26:25,519 --> 00:26:27,400 have to play the skeptic here, please do Is this 541 00:26:27,519 --> 00:26:29,960 just billionaire hubris? Are we just building a high tech 542 00:26:30,079 --> 00:26:33,680 Jurassic Park vanity project? Dropping a herd of mammoth hybrids 543 00:26:33,720 --> 00:26:37,240 into the modern Arctic tundra, an ecosystem that has adapted 544 00:26:37,279 --> 00:26:40,079 without them for a millennia seems incredibly reckless. 545 00:26:40,119 --> 00:26:43,079 Speaker 2: The ecological concerns are completely valid, but framing it as 546 00:26:43,079 --> 00:26:47,240 a vanity project misses the underlying technological achievement. Colossal's work 547 00:26:47,279 --> 00:26:48,839 is not ultimately about the mammoth. 548 00:26:48,920 --> 00:26:49,200 Speaker 1: It's not. 549 00:26:49,480 --> 00:26:53,200 Speaker 2: No, it is about building the multiplex genomic engineering tools 550 00:26:53,200 --> 00:26:57,079 required to edit forty five genes simultaneously into a living embryo. 551 00:26:57,519 --> 00:26:59,319 If you can do that with an elephant, you can 552 00:26:59,359 --> 00:27:02,960 apply those exact same tools to current endangered species. 553 00:27:03,039 --> 00:27:03,640 Speaker 1: Oh I see. 554 00:27:03,680 --> 00:27:07,119 Speaker 2: You can engineer coral reefs to withstand higher ocean temperatures. 555 00:27:07,599 --> 00:27:11,640 You can engineer agricultural crops to survive extreme drought. The 556 00:27:11,680 --> 00:27:14,920 mammoth is the moonshot project designed to attract capital and 557 00:27:14,960 --> 00:27:18,559 push the boundaries of genetic engineering, a proof of concept exactly. 558 00:27:18,839 --> 00:27:21,599 The real prize is the ability to program climate resilience 559 00:27:21,680 --> 00:27:22,960 into the biosphere. 560 00:27:23,079 --> 00:27:26,079 Speaker 1: Okay, that makes the ten billion dollar valuation make a 561 00:27:26,079 --> 00:27:29,119 lot more sense. But the ability to edit life isn't 562 00:27:29,160 --> 00:27:32,640 stopping at animals. It's moving directly into human medicine. Yes, 563 00:27:32,640 --> 00:27:35,480 it is the MIT data on base edited babies and 564 00:27:35,519 --> 00:27:38,839 precision gene editing. Breakthrough Number three is where this gets 565 00:27:38,880 --> 00:27:43,519 intensely personal. In May twenty twenty five, researchers use Crisper 566 00:27:43,559 --> 00:27:46,400 technology to treat a newborn with a severe condition called 567 00:27:46,480 --> 00:27:48,599 carbamoil phosphate synthatase IE. 568 00:27:48,480 --> 00:27:50,279 Speaker 2: Deficiency within six months. 569 00:27:50,480 --> 00:27:54,880 Speaker 1: Yes, they designed, manufactured, and delivered a bespoke genetic therapy 570 00:27:54,920 --> 00:27:58,839 within six months. By late twenty twenty four, fifty centers 571 00:27:58,839 --> 00:28:02,240 were offering Crisper their bees to over fifty patients. We 572 00:28:02,319 --> 00:28:04,799 are directly correcting typos in human DNA. 573 00:28:05,160 --> 00:28:08,759 Speaker 2: The acceleration of this timeline is breathtaking. We are moving 574 00:28:08,759 --> 00:28:12,440 from a paradigm of managing symptoms with lifelong pharmacology to 575 00:28:12,519 --> 00:28:16,480 a paradigm of definitive, single treatment cures. You send in 576 00:28:16,519 --> 00:28:19,839 a microscopic mechanism to find the exact genetic error, cut 577 00:28:19,839 --> 00:28:21,839 it out and paste in the correct sequence. 578 00:28:22,000 --> 00:28:23,440 Speaker 1: Direct DNA correction. 579 00:28:23,279 --> 00:28:25,920 Speaker 2: But the MIT data indicates that as these tools become 580 00:28:26,000 --> 00:28:29,240 safer and more precise, the intervention window will move earlier. 581 00:28:29,279 --> 00:28:32,880 We will move from treating newborns to editing embryos before birth. 582 00:28:32,920 --> 00:28:36,079 Speaker 1: And here's where we hit the ethical tightrope Eradicating horrific 583 00:28:36,160 --> 00:28:40,160 inherited genetic diseases is a monumental triumph for humanity. No 584 00:28:40,200 --> 00:28:42,440 one debates that. But if we can edit out a 585 00:28:42,480 --> 00:28:44,039 debilitating disease. 586 00:28:43,839 --> 00:28:45,839 Speaker 2: What else can we edit designer babies? 587 00:28:46,279 --> 00:28:50,240 Speaker 1: Exactly? If you have the tools to alter the biological 588 00:28:50,240 --> 00:28:53,039 blueprint of a human being before they are born, who 589 00:28:53,039 --> 00:28:56,240 decides where the line is between a disease and a trait? 590 00:28:57,119 --> 00:28:59,640 We are stepping into a realm where we are actively 591 00:28:59,640 --> 00:29:03,079 guiding our own evolution, and I'm not entirely convinced our 592 00:29:03,119 --> 00:29:06,480 ethical frameworks have caught up to our engineering capabilities. 593 00:29:06,519 --> 00:29:10,400 Speaker 2: They definitely haven't. The magnitude of that human intervention cannot 594 00:29:10,400 --> 00:29:13,720 be overstated. We are navigating an errow where the software, 595 00:29:13,839 --> 00:29:16,480 the energy grid, and our very biology are all becoming 596 00:29:16,519 --> 00:29:19,640 reprogrammable simultaneously to everything, all at once, And as we 597 00:29:19,680 --> 00:29:23,319 rewrite our biology, we are also physically remapping how we 598 00:29:23,359 --> 00:29:25,319 move through the world. Which brings us to the final 599 00:29:25,359 --> 00:29:28,880 cluster of breakthroughs, the physical and digital infrastructure of the 600 00:29:28,920 --> 00:29:33,039 late twenty twenties. Let's look at urban mobility with EVTs, 601 00:29:33,359 --> 00:29:36,000 electric vertical takeoff and landing vehicles. 602 00:29:35,640 --> 00:29:40,160 Speaker 1: Air taxis breakthrough number twelve. For twenty years, I've seen 603 00:29:40,200 --> 00:29:43,799 the glossy three D renders of flying taxis in tech magazines, 604 00:29:44,039 --> 00:29:46,720 and it always felt like vaporware, just concept art. 605 00:29:46,799 --> 00:29:47,920 Speaker 2: But it's not anymore. 606 00:29:48,200 --> 00:29:50,720 Speaker 1: The MIT report makes it clear that we've crossed the 607 00:29:50,759 --> 00:29:55,839 threshold from concept art to rigorous certified flight testing. Jobey 608 00:29:55,920 --> 00:29:59,000 Aviation conducted over eight hundred and fifty flights in twenty 609 00:29:59,039 --> 00:30:01,559 twenty five, logging fifty thousand miles. 610 00:30:02,079 --> 00:30:06,039 Speaker 2: Beta technologies aircrafts surpassed one hundred thousand nautical miles. 611 00:30:06,480 --> 00:30:09,640 Speaker 1: Archer's Midnight aircraft hit ten thousand feet on a fifty 612 00:30:09,680 --> 00:30:13,160 five mile flight. They are flown at the Osaka World Expo. 613 00:30:13,279 --> 00:30:16,480 They are flying in the UAE. This is real world 614 00:30:16,519 --> 00:30:17,480 performance data. 615 00:30:17,559 --> 00:30:22,079 Speaker 2: The engineering is proven. The next phase is certification and infrastructure. 616 00:30:22,519 --> 00:30:25,519 What fascinates me is how urban air mobility will alter 617 00:30:25,599 --> 00:30:28,079 the geometry of real estate and city planning. 618 00:30:28,359 --> 00:30:29,640 Speaker 1: How so well. 619 00:30:29,799 --> 00:30:32,640 Speaker 2: Real estate values have always been fundamentally tethered to commute 620 00:30:32,680 --> 00:30:37,119 times via terrestrial infrastructure, roads, bridges, and rail lines. If 621 00:30:37,160 --> 00:30:40,119 you can traverse a sprawling metropolitan area in a ten 622 00:30:40,160 --> 00:30:43,480 minute aerial hop by passing all ground traffic, the radius 623 00:30:43,519 --> 00:30:45,799 of where a workforce can live while still participating in 624 00:30:45,839 --> 00:30:48,039 a city center's economy expands dramatically. 625 00:30:48,079 --> 00:30:49,880 Speaker 1: You could live in the mountains and commute to downtown 626 00:30:49,960 --> 00:30:50,559 in minutes. 627 00:30:50,680 --> 00:30:54,319 Speaker 2: Exactly, infrastructure budgets will shift from highway expansion to a 628 00:30:54,359 --> 00:30:56,079 network of distributed vertiports. 629 00:30:56,519 --> 00:30:59,599 Speaker 1: Just imagine the reclaim time for a second. Instead of 630 00:30:59,599 --> 00:31:01,480 losing ten ten hours a week staring at the bumper 631 00:31:01,519 --> 00:31:03,200 of the car in front of you, you take a 632 00:31:03,279 --> 00:31:06,599 quiet electric flight over the gridlock. And once we conquer 633 00:31:06,680 --> 00:31:10,160 the airspace above our cities, the next logistical frontier is 634 00:31:10,279 --> 00:31:11,079 low Earth orbit. 635 00:31:11,200 --> 00:31:13,039 Speaker 2: Break Through number eleven. 636 00:31:12,680 --> 00:31:16,160 Speaker 1: Commercial space stations. NASA awarded four hundred and fifteen point 637 00:31:16,160 --> 00:31:20,000 six million dollars to private companies to develop commercial space stations. 638 00:31:20,480 --> 00:31:23,400 The Axiumission four in twenty twenty five ran over sixty 639 00:31:23,400 --> 00:31:26,920 experiments from thirty one different countries in just eighteen days. 640 00:31:27,440 --> 00:31:30,799 We are transitioning from space as a government exploration zone 641 00:31:31,000 --> 00:31:33,079 to space as a commercial industrial park. 642 00:31:33,759 --> 00:31:38,440 Speaker 2: History demonstrates a consistent pattern here. Infrastructure always precedes industry. 643 00:31:38,799 --> 00:31:41,000 You lay the rail lines before you build the factories. 644 00:31:41,240 --> 00:31:43,880 We are currently laying the logistical rail lines to low 645 00:31:43,920 --> 00:31:44,440 Earth orbit. 646 00:31:44,680 --> 00:31:45,839 Speaker 1: What are those factories built? 647 00:31:46,079 --> 00:31:49,680 Speaker 2: Private space stations represent a paradigm shift in manufacturing. When 648 00:31:49,720 --> 00:31:52,519 you remove the variable of gravity, you can create perfect 649 00:31:52,519 --> 00:31:58,640 molecular structures. Pharmaceutical crystallization, advanced fiber optics, and even synthetic 650 00:31:58,759 --> 00:32:02,039 organs can be manufactured microgravity with a level of precision 651 00:32:02,279 --> 00:32:04,839 that is physically impossible on Earth's surface. 652 00:32:05,039 --> 00:32:08,359 Speaker 1: The idea of buying a high end consumer product with 653 00:32:08,480 --> 00:32:12,559 a made in orbit label is wild, but it's happening. 654 00:32:12,799 --> 00:32:16,079 But here is the terrifying reality check that underscores everything 655 00:32:16,079 --> 00:32:18,799 we've talked about today, the security flaw. Yes, we are 656 00:32:18,839 --> 00:32:23,880 building an autonomous, AI driven, nuclear powered society, expanding into 657 00:32:23,960 --> 00:32:27,960 orbital manufacturing and genetic engineering, and almost all of the 658 00:32:28,000 --> 00:32:32,519 digital communication securing this entire global apparatus is protected by 659 00:32:32,599 --> 00:32:34,759 cryptographic math from the nineteen nineties. 660 00:32:34,839 --> 00:32:38,240 Speaker 2: It is the ultimate systemic vulnerability that brings us. 661 00:32:38,200 --> 00:32:41,839 Speaker 1: To our final breakthrough number five, quantum secure cryptography. 662 00:32:42,000 --> 00:32:46,039 Speaker 2: The entire digital economy, Every bank transfer every military communication, 663 00:32:46,400 --> 00:32:49,079 the telemetry data for those of Vichikol. The control systems 664 00:32:49,079 --> 00:32:52,559 for the SMRs relies on encryption standards like RSA twenty 665 00:32:52,599 --> 00:32:55,880 forty eight. Those standards are based on mathematical problems that 666 00:32:55,920 --> 00:33:00,519 would take traditional computers millions of years to solve. Computing 667 00:33:00,559 --> 00:33:04,759 utilizes a completely different model of processing. In twenty twenty five, 668 00:33:04,839 --> 00:33:09,279 researchers demonstrated algorithmic improvements that made breaking RSA twenty forty 669 00:33:09,319 --> 00:33:12,440 eight approximately twenty times easier than previously. 670 00:33:12,000 --> 00:33:13,359 Speaker 1: Modeled twenty times easier. 671 00:33:13,400 --> 00:33:15,640 Speaker 2: The margin of safety is evaporating rapidly. 672 00:33:15,759 --> 00:33:18,319 Speaker 1: It's like a silent digital y two K, but this 673 00:33:18,440 --> 00:33:21,880 time the threat is very real. If quantum decryption matures 674 00:33:21,880 --> 00:33:25,680 before we upgrade our security protocols, defense networks are compromised, 675 00:33:26,079 --> 00:33:29,440 the global financial system becomes completely transparent to bad actors. 676 00:33:29,839 --> 00:33:31,480 Commerce just stops. 677 00:33:31,599 --> 00:33:33,920 Speaker 2: It's an existential threat to the digital age. 678 00:33:34,039 --> 00:33:35,960 Speaker 1: The good news in the MIT data is that the 679 00:33:35,960 --> 00:33:38,440 brightest minds are racing to swap the locks before the 680 00:33:38,519 --> 00:33:41,559 quantum lock picks are finished. Over fifty percent of human 681 00:33:41,599 --> 00:33:45,319 web traffic is already protected by post quantum key systems support. 682 00:33:45,359 --> 00:33:48,079 Among the top one hundred thousand Internet domains jump from 683 00:33:48,079 --> 00:33:51,279 twenty eight percent to thirty nine percent. Regulators are drawing 684 00:33:51,279 --> 00:33:54,200 lines in the sand, setting strict migration deadlines for twenty 685 00:33:54,200 --> 00:33:56,960 thirty and twenty thirty five. We are actively rebuilding the 686 00:33:57,000 --> 00:33:59,079 vault doors while the bank is still open for business. 687 00:33:59,240 --> 00:34:01,559 Speaker 2: If you synth the size the entirety of this landscape, 688 00:34:01,559 --> 00:34:03,759 you realize we are not dealing with a list of 689 00:34:03,799 --> 00:34:04,920 isolated inventions. 690 00:34:05,039 --> 00:34:06,400 Speaker 1: No, they're all deeply connected. 691 00:34:06,599 --> 00:34:10,880 Speaker 2: We are entering an era of cascading breakthroughs. Agentic AI 692 00:34:10,960 --> 00:34:15,679 systems will utilize quantum secure cryptography to securely orchestrate supply 693 00:34:15,800 --> 00:34:19,440 chains of autonomous EVE tools. Those systems will be powered 694 00:34:19,440 --> 00:34:23,400 by a grid transformed by perofskite, solar aluminum fuels, and 695 00:34:23,480 --> 00:34:27,480 small modular reactors. The infinite compute will map the mechanistic 696 00:34:27,519 --> 00:34:31,320 interpretability of the AI itself, while simultaneously allowing us to 697 00:34:31,360 --> 00:34:35,840 engineer our biological resilience. Each technology acts as a multiplier 698 00:34:35,840 --> 00:34:39,039 for the others, accelerating the pace of change exponentially. 699 00:34:39,159 --> 00:34:43,239 Speaker 1: It is a breath taking, almost overwhelming web of human ingenuity. 700 00:34:44,039 --> 00:34:45,840 But as we look at this world, we are building 701 00:34:45,840 --> 00:34:49,280 a world where software acts autonomously. Energy is infinite, and 702 00:34:49,360 --> 00:34:52,800 biology is reprogrammable. I keep coming back to a single 703 00:34:52,840 --> 00:34:55,760 provocative thought that as we gain the ability to control 704 00:34:55,840 --> 00:34:59,440 the literal source code of everything, we are systematically engineering 705 00:34:59,440 --> 00:35:03,639 out friction. We are engineering out inefficiency. But what happens 706 00:35:03,639 --> 00:35:06,320 to the concept of human error when a system is 707 00:35:06,480 --> 00:35:11,639 entirely automated, perfectly reprogrammable, and frictionless. Who gets to decide 708 00:35:11,679 --> 00:35:14,239 the blueprint? If we strip out all the mistakes, do 709 00:35:14,320 --> 00:35:18,519 we also risk stripping out the beautiful, unpredictable, messy mutations 710 00:35:18,559 --> 00:35:20,280 that actually drive human progress. 711 00:35:20,360 --> 00:35:22,679 Speaker 2: That is the ultimate question of the twenty first century. 712 00:35:22,960 --> 00:35:26,119 The tools we have discussed today are morally neutral. They 713 00:35:26,159 --> 00:35:29,679 are levers of immense power. The intent behind the blueprint, 714 00:35:29,760 --> 00:35:32,519 the wisdom with which we deploy these systems will determine 715 00:35:32,519 --> 00:35:35,800 whether this era of abundance serves humanity or simply manages it. 716 00:35:36,039 --> 00:35:38,320 Speaker 1: And that is exactly where we want to bring you, 717 00:35:38,559 --> 00:35:42,519 the listener, into the conversation. We have covered a massive 718 00:35:42,559 --> 00:35:46,079 amount of ground today on thrilling threads, from aluminum batteries 719 00:35:46,079 --> 00:35:49,559 to mammoth hybrids. Which of these fifteen breakthroughs are you 720 00:35:49,719 --> 00:35:51,960 most excited to see become a part of your daily 721 00:35:52,000 --> 00:35:55,960 life and honestly, which one secretly keeps you up at night? 722 00:35:56,440 --> 00:35:58,440 Are you ready to hand over your calendar and your 723 00:35:58,480 --> 00:36:01,920 credit card to an AGENTIC or are you just holding 724 00:36:01,920 --> 00:36:03,559 out for the day. You can skip the commute and 725 00:36:03,639 --> 00:36:06,360 electric air taxi. Drop a comment below and let us 726 00:36:06,360 --> 00:36:08,800 know your stand. Thank you so much for joining us 727 00:36:08,840 --> 00:36:12,440 for this deep exploration. Remember to stay curious, keep asking 728 00:36:12,440 --> 00:36:15,199 the hard questions, and keep pulling at the threads of tomorrow.