1 00:00:00,840 --> 00:00:04,770 This isn't for everyone. It's for the ones who know the 2 00:00:04,920 --> 00:00:07,590 old system is broken and are ready to build something that 3 00:00:07,680 --> 00:00:22,400 lasts. Welcome to the Build with Bitcoin podcast. We're co-host 4 00:00:22,490 --> 00:00:25,430 Iraiden Lynne, and today we're doing a deep dive with Early 5 00:00:25,550 --> 00:00:28,370 Writers partner Liam Nelson. We're going to get into your 6 00:00:28,850 --> 00:00:31,760 research report. We'll be unpacking everything called the 7 00:00:31,790 --> 00:00:35,240 Singularity Stack. But first of all, welcome to the podcast, and 8 00:00:35,450 --> 00:00:36,110 thanks for joining us. 9 00:00:36,230 --> 00:00:38,720 Thanks for having me. First time, long time. Really 10 00:00:38,750 --> 00:00:41,320 appreciate the show and all the knowledge that you guys have put 11 00:00:41,410 --> 00:00:42,910 out over the years, so thanks for having me. 12 00:00:43,059 --> 00:00:45,459 We we appreciate you guys, and early congrats on the 13 00:00:45,549 --> 00:00:47,739 report because it's a fascinating research project 14 00:00:47,769 --> 00:00:52,299 that you guys put out. And before that, let's maybe get to 15 00:00:52,359 --> 00:00:54,819 know you a little bit. I mean, what can you share about your 16 00:00:54,879 --> 00:00:57,279 background, Liam, and what brought you to Bitcoin? 17 00:00:57,399 --> 00:01:01,019 Yeah, for sure. Just really quick background on 18 00:01:01,019 --> 00:01:04,859 was finance major at Georgetown Business School, and 19 00:01:05,549 --> 00:01:09,449 then started off my career at Point 72. For those who don't 20 00:01:09,479 --> 00:01:13,589 know, it's founded by Steve Cohen, and they are about a $40 21 00:01:13,649 --> 00:01:17,579 billion long short hedge fund, primarily trading U.S. equities. 22 00:01:18,179 --> 00:01:22,579 And I was there from 2019 to 2022, and the time there was 23 00:01:22,669 --> 00:01:26,449 particularly interesting because I was able to join for all the 24 00:01:26,599 --> 00:01:29,719 issues related to tariffs, and then was covering a lot of the 25 00:01:29,899 --> 00:01:34,309 consumer and industrial sectors during COVID too, and so was 26 00:01:34,429 --> 00:01:36,979 able to to cover companies like Walmart, Target, all the 27 00:01:37,069 --> 00:01:40,449 consumer goods. Is there was issues related to toilet paper 28 00:01:40,539 --> 00:01:44,109 running out, and and then got to be able to see everything from 29 00:01:44,229 --> 00:01:48,939 the run up after that too. As there, I was doing deep dives on 30 00:01:49,329 --> 00:01:52,269 on businesses and companies, and seeing the fundamentals be 31 00:01:52,509 --> 00:01:56,109 completely dislocated from the actual valuations of the stock 32 00:01:56,139 --> 00:02:00,419 prices. And so during that time, both both myself trying to 33 00:02:00,419 --> 00:02:02,999 understand what was going on from an individual level, as 34 00:02:03,059 --> 00:02:06,629 well as the firm, really was going deep into the macro side 35 00:02:06,659 --> 00:02:09,899 of things too, and obviously trying to understand what's 36 00:02:09,959 --> 00:02:13,949 going on with the Fed, monetary policy, etc. And that's a lot of 37 00:02:14,069 --> 00:02:17,879 what led me to find Bitcoin during my time there. Point 72 38 00:02:18,059 --> 00:02:21,739 had an average hold period of only 33 days, so incredibly 39 00:02:21,889 --> 00:02:27,229 short and very thesis in momentum and like event-driven 40 00:02:27,349 --> 00:02:31,849 firm for holding stocks, and so it was a fascinating time to be 41 00:02:32,029 --> 00:02:34,699 there, where there was no shortage of events going on 42 00:02:34,999 --> 00:02:39,319 around that time, and we actually worked and funded 43 00:02:39,829 --> 00:02:43,119 Melvin Capital on the wrong side of the GameStop trade, and so 44 00:02:43,899 --> 00:02:48,339 we-I wasn't personally in there, but we, as a firm, we were 45 00:02:48,399 --> 00:02:50,829 pretty close, and so that was just the point that I was like, 46 00:02:51,069 --> 00:02:53,769 "What the hell is going on with the entire financial system? 47 00:02:53,859 --> 00:02:57,219 There's some something else going on at play, and and really 48 00:02:57,309 --> 00:03:01,199 found Bitcoin deeply tried to you know spend all of my waking 49 00:03:01,379 --> 00:03:04,469 hours outside of the firm to go deep there and understand what 50 00:03:04,469 --> 00:03:07,739 was going on, and and as we all kind of come out the other side, 51 00:03:07,799 --> 00:03:11,399 understanding where this all goes in the long term. That was 52 00:03:11,489 --> 00:03:15,959 my kind of moment there. I spent a few years after that in 53 00:03:16,049 --> 00:03:19,619 private equity at a firm called Kobe Capital, and then joined 54 00:03:19,769 --> 00:03:23,929 Early Riders about two years ago now, had read their white paper 55 00:03:24,379 --> 00:03:27,799 around Bitcoin being the hurdle rate and how it aligns 56 00:03:27,859 --> 00:03:33,019 incentives from LPs, GPs, and founders all trying to make more 57 00:03:33,109 --> 00:03:35,449 Bitcoin and using that as a hurdle rate. And it just made 58 00:03:35,539 --> 00:03:38,419 too much sense to me, and thought that all other financial 59 00:03:38,479 --> 00:03:40,599 firms, whether it be private equity, et cetera, would be 60 00:03:41,409 --> 00:03:44,559 either adopt this model eventually or be competed away, 61 00:03:44,949 --> 00:03:47,739 and so yeah, that's a little bit of my background and how we got 62 00:03:47,799 --> 00:03:48,969 here. I 63 00:03:49,120 --> 00:03:51,430 love it. You you really went from extreme high 64 00:03:51,880 --> 00:03:55,000 time preference to low time preference in terms of your 65 00:03:55,420 --> 00:03:58,180 investment theories and and outlooks. And and we've had 66 00:03:58,690 --> 00:04:02,940 Michael Tanguma on our podcast before, and early writers, and 67 00:04:03,060 --> 00:04:05,700 we love the story. I mean, we we couldn't agree more with you 68 00:04:05,760 --> 00:04:07,380 that we think that that's ultimately the way, the 69 00:04:07,650 --> 00:04:10,950 direction everything needs to flow. And I want to talk a 70 00:04:10,950 --> 00:04:14,400 little bit more about your role and and how you came to write 71 00:04:14,460 --> 00:04:17,130 this report, the Singularity Stack. So you you're really 72 00:04:17,370 --> 00:04:20,460 focused as a partner in in the firm on due diligence and 73 00:04:20,490 --> 00:04:24,260 portfolio operations. How did how did you come to create this 74 00:04:24,380 --> 00:04:27,140 research report? I mean, and I can't wait to dive into it 75 00:04:27,140 --> 00:04:30,470 because I just find it fascinating. But did you create 76 00:04:30,530 --> 00:04:32,750 it as a kind of a framework for how you're looking at the 77 00:04:32,810 --> 00:04:36,080 companies that come across your desk, or or how did how did that 78 00:04:36,710 --> 00:04:36,980 arise? 79 00:04:37,160 --> 00:04:41,380 Yeah, honestly, we try to pay attention to a lot of 80 00:04:41,500 --> 00:04:45,940 interesting signals out there, and I right now AI is obviously 81 00:04:45,940 --> 00:04:49,060 the most interesting thing in the room. There was a chart 82 00:04:49,090 --> 00:04:53,050 going around yesterday too that Open Router put out around token 83 00:04:53,110 --> 00:04:57,970 usage, and it was over the past two years, it's up over 25,000% 84 00:04:58,810 --> 00:05:01,680 and that's obviously just. Astounding! It's really showing 85 00:05:01,740 --> 00:05:06,660 just the amount of adoption that this that these tools are 86 00:05:06,720 --> 00:05:09,810 getting, and it's really fascinating too because, like, 87 00:05:09,870 --> 00:05:13,560 as an investor and and somebody who is trying to understand 88 00:05:13,620 --> 00:05:16,470 where the world is moving, I saw a lot of similarities between 89 00:05:16,560 --> 00:05:20,130 like what Jeff Bezos was doing when he was at D Shaw and and 90 00:05:20,190 --> 00:05:25,910 saw the internet grow. I think it was 2300 time percent year 91 00:05:25,970 --> 00:05:28,580 over year, which doesn't necessarily happen. And that's 92 00:05:28,610 --> 00:05:31,940 something that's free to use for any individual. Whereas you 93 00:05:31,970 --> 00:05:35,540 know, obviously, tokens on open router have a cost associated 94 00:05:35,600 --> 00:05:41,380 with them. And so obviously, out of that, he built Amazon and you 95 00:05:41,410 --> 00:05:44,950 know, selling books online at first doesn't sound very sexy, 96 00:05:45,490 --> 00:05:48,820 but we understood kind of by being on the frontier that it 97 00:05:48,820 --> 00:05:51,640 wasn't just going to be books; it was going to be more than 98 00:05:51,790 --> 00:05:53,680 that. They were going to sell everything, and obviously got 99 00:05:53,710 --> 00:05:57,730 into data centers out of that, and have try like built it into 100 00:05:57,760 --> 00:06:00,930 a massive company. And so I think that you know, just 101 00:06:00,960 --> 00:06:03,540 myself, and and I think a lot of people out there really are 102 00:06:03,600 --> 00:06:06,090 understand are trying to understand what are the second, 103 00:06:06,150 --> 00:06:09,450 third, fourth order implications of AI as the cost of 104 00:06:09,480 --> 00:06:15,540 intelligence goes near to zero, and what sort of businesses can 105 00:06:15,690 --> 00:06:18,330 be built out of that. What does that mean for my role today and 106 00:06:18,420 --> 00:06:20,480 in the future? And so that's kind of a little bit of the 107 00:06:20,600 --> 00:06:24,380 impetus of why I tried to do it because I think that it does 108 00:06:24,440 --> 00:06:27,500 have just profound implications for pretty much every part of 109 00:06:27,500 --> 00:06:29,870 the economy. Yeah, and I imagine it it clicked with the 110 00:06:30,080 --> 00:06:32,810 early writers team, the other partners right away, and 111 00:06:33,140 --> 00:06:36,020 kind of gave you the green light to to work on this. Well, I 112 00:06:36,080 --> 00:06:40,220 mean, before we we get into the framework, Liam, I want to just 113 00:06:40,280 --> 00:06:44,770 make sure I don't miss on asking you. Sometimes we, you know, we 114 00:06:44,860 --> 00:06:47,020 kind of go through our background and and keep it 115 00:06:47,020 --> 00:06:50,530 concise, and the transitions sound pretty smooth. But for 116 00:06:50,650 --> 00:06:53,230 others who are maybe going through, you know, similar 117 00:06:53,350 --> 00:06:56,680 career transitions or desiring to do so, I want to ask this 118 00:06:56,740 --> 00:07:01,500 question. I mean, from point 72 to working at a at a Bitcoin 119 00:07:02,550 --> 00:07:07,380 Bitcoin firm, it's not that easy of a transition necessarily. So, 120 00:07:07,830 --> 00:07:10,020 what helped you along the way? 121 00:07:10,230 --> 00:07:12,360 Yeah, that's a great question, and really appreciate 122 00:07:12,570 --> 00:07:17,700 it. I guess I was coming at it from the angle that there was a 123 00:07:17,790 --> 00:07:20,610 massive information asymmetry that's publicly available, 124 00:07:21,120 --> 00:07:24,650 whereas you know you guys put out a ton of great research and 125 00:07:25,010 --> 00:07:27,950 really make the information available about why Bitcoin's so 126 00:07:28,010 --> 00:07:31,010 important, etc. But there are so many folks out there that are 127 00:07:31,310 --> 00:07:36,200 you know even J.P. Morgan or Jamie Dimon from J.P. Morgan 128 00:07:36,590 --> 00:07:39,530 says like Satoshi can just come back and inflate the Bitcoin 129 00:07:39,680 --> 00:07:44,260 supply, and so with that, like I understood just by speaking with 130 00:07:44,290 --> 00:07:46,990 a bunch of folks that there was a lot of information asymmetry, 131 00:07:47,050 --> 00:07:50,620 and you know, eventually they would come around to why Bitcoin 132 00:07:50,680 --> 00:07:54,400 is valuable, and you know, would would need to allocate to it 133 00:07:54,430 --> 00:07:58,480 later on, just because economic forces would eventually get them 134 00:07:58,540 --> 00:08:01,530 there, because you know they would see their savings diminish 135 00:08:01,650 --> 00:08:05,340 over time in real value if they didn't allocate to Bitcoin, and 136 00:08:05,400 --> 00:08:09,450 so with that, I I understood you know it's it makes sense to 137 00:08:09,540 --> 00:08:13,380 allocate to the asset itself, and then I tried to understand 138 00:08:13,440 --> 00:08:17,370 well there could be other implications here too if you 139 00:08:17,430 --> 00:08:20,900 know the monetary system is inflating 7% a year, like 140 00:08:20,930 --> 00:08:24,770 there's going to be better information that can teach 141 00:08:24,860 --> 00:08:29,150 people about that, whether it's the internet, podcasts, AI, and 142 00:08:29,270 --> 00:08:31,700 there are going to be other implications for that too. And 143 00:08:31,760 --> 00:08:35,780 so, one of them that I thought about was there are going to be 144 00:08:36,350 --> 00:08:40,040 companies that will, you know, benefit from a lot more folks 145 00:08:40,100 --> 00:08:43,840 who trade Bitcoin? Who want to custody Bitcoin, lend against 146 00:08:44,560 --> 00:08:49,780 it, etc. There will be like an entire changing of financial 147 00:08:49,870 --> 00:08:52,600 services around that. And I actually thought about it too, 148 00:08:52,780 --> 00:08:55,570 and and I was like, well, if there are X amount of people 149 00:08:55,690 --> 00:08:59,290 today that own Bitcoin and Bitcoin's going to go up Y 150 00:08:59,440 --> 00:09:04,560 amount, like the amount of new, like new people trading Bitcoin 151 00:09:04,650 --> 00:09:07,770 and and financial services around it will actually likely 152 00:09:07,920 --> 00:09:12,180 increase faster than just Bitcoin itself, and so that that 153 00:09:12,180 --> 00:09:14,820 was kind of a moment that clicked for me that tried to 154 00:09:14,820 --> 00:09:18,780 understand like, well, if you back the right founders who are 155 00:09:18,870 --> 00:09:24,140 building interesting categories in here, then there's a massive 156 00:09:24,320 --> 00:09:28,100 opportunity for growth, and it's a really big investable category 157 00:09:28,250 --> 00:09:31,190 to back the infrastructure. To date, we've seen a lot of 158 00:09:31,640 --> 00:09:36,470 distractions with digital asset companies. I think you know 159 00:09:36,590 --> 00:09:40,430 focus on on like meme coins and whatever from like Coinbase's of 160 00:09:40,430 --> 00:09:44,110 the world, but I think they've obviously financially still done 161 00:09:44,230 --> 00:09:47,860 extremely well. I think that if you really develop 162 00:09:48,310 --> 00:09:51,820 differentiated products and see that growth, like it's going to 163 00:09:51,850 --> 00:09:55,660 be an extremely attractive opportunity. And so I had that 164 00:09:55,750 --> 00:10:01,050 in mind, and eventually, just you know. Spoke with enough 165 00:10:01,260 --> 00:10:04,410 companies out there in the space to understood that that was the 166 00:10:04,530 --> 00:10:08,160 right time. The consumers were kind of waking up, and the 167 00:10:08,310 --> 00:10:13,350 infrastructure wasn't just it was maturing at the right time. 168 00:10:13,530 --> 00:10:16,140 That there were more folks who were ready to come into the 169 00:10:16,170 --> 00:10:19,680 space, and so, but obviously not not yet developed enough for 170 00:10:19,770 --> 00:10:22,910 everybody, and so that's why I kind of made the the leap to go 171 00:10:22,970 --> 00:10:25,700 to early riders to help invest in other infrastructure 172 00:10:25,790 --> 00:10:26,630 companies in the space. 173 00:10:27,470 --> 00:10:29,690 Well, I think this is a good time to kind of 174 00:10:29,750 --> 00:10:33,290 get into the analogy of building the infrastructure for Bitcoin 175 00:10:33,410 --> 00:10:36,860 and what Bitcoin financial rails will look like, and how you 176 00:10:36,950 --> 00:10:40,010 start your report, the Singularity stack, with kind of 177 00:10:40,100 --> 00:10:44,110 the the a look back at at electricity and electrical 178 00:10:44,140 --> 00:10:46,870 lighting. So can you can you start us out by kind of because 179 00:10:46,900 --> 00:10:50,350 I do think that there is a lack of understanding about how 180 00:10:50,470 --> 00:10:55,210 important Bitcoin is for future financial services. So maybe you 181 00:10:55,270 --> 00:10:58,810 can can bring us that analogy that might help people that 182 00:10:58,900 --> 00:11:02,550 aren't understanding that Bitcoin is so much more than an 183 00:11:02,730 --> 00:11:05,400 asset that trades, but it's really a foundational protocol. 184 00:11:05,580 --> 00:11:08,970 Yeah, for sure. I guess taking step back. Yeah, the the 185 00:11:09,810 --> 00:11:14,370 light. We we started off the paper about the AI singularity 186 00:11:14,520 --> 00:11:17,670 with the analogy to light because I think that over time 187 00:11:17,820 --> 00:11:21,300 intelligence is something that's going to be essentially 188 00:11:21,360 --> 00:11:24,740 available for everybody, and going to be so cheap that it's 189 00:11:24,770 --> 00:11:28,370 going to be very difficult to even meter it. And so, back in 190 00:11:28,370 --> 00:11:31,970 the day, like 200 years ago, it would cost three hours of 191 00:11:32,810 --> 00:11:37,010 somebody's time in order to get an hour of light, and that was 192 00:11:37,100 --> 00:11:41,770 just because there it was craft good. Like you had to actually 193 00:11:41,860 --> 00:11:47,680 create, like get whale oil, etc. and create candles yourself. 194 00:11:47,890 --> 00:11:52,240 Rather, and there wasn't really any use to being up late at 195 00:11:52,300 --> 00:11:55,870 night because everybody was farmers, and then they're like, 196 00:11:56,140 --> 00:11:59,410 "What? What can you really even do after light? It's just read a 197 00:11:59,440 --> 00:12:03,720 book or whatever. But today it costs like less than a 10th of a 198 00:12:03,810 --> 00:12:07,200 second in order to have an hour of light because we've built up 199 00:12:07,350 --> 00:12:11,430 all the infrastructure around it, and so I think that 200 00:12:11,550 --> 00:12:13,830 obviously had massive implications that I think we're 201 00:12:13,980 --> 00:12:20,130 about to see now as it relates to the first and like first and 202 00:12:20,190 --> 00:12:23,210 second industrial revolution and how there was then a build out 203 00:12:23,390 --> 00:12:28,520 of 24/7 manufacturing and supply chain that will go around the 204 00:12:28,550 --> 00:12:31,640 world, and right now we're starting to see the same amount 205 00:12:31,640 --> 00:12:36,110 of it with AI and intelligence coming from a craft good that 206 00:12:36,710 --> 00:12:40,450 you know you have to sit down and think about for a certain 207 00:12:40,480 --> 00:12:44,440 amount of time versus something that's just on hand callable and 208 00:12:44,830 --> 00:12:48,940 can be, you know, metered by the second. And so we're starting to 209 00:12:49,030 --> 00:12:53,800 change the way that we actually pay for goods, and obviously 210 00:12:53,830 --> 00:12:57,250 intelligence is one of the biggest ones there, and that's 211 00:12:57,280 --> 00:13:02,490 going to have implications for not just the the way that the 212 00:13:02,550 --> 00:13:07,320 economy works, but also how it ends up in which payment method 213 00:13:07,680 --> 00:13:09,660 is ultimately used in the long term. 214 00:13:09,750 --> 00:13:13,620 So let's kind of connect the dots here, Liam. The report 215 00:13:13,650 --> 00:13:17,520 lays out nine layers, starting with with raw electricity. Of 216 00:13:17,640 --> 00:13:21,200 course, this all eventually ties to value and money. Can you take 217 00:13:21,200 --> 00:13:23,900 us through what the stack is? What what the foundation of the 218 00:13:23,900 --> 00:13:24,050 stack 219 00:13:24,410 --> 00:13:29,990 is? Yeah, real quick. It's essentially everything is 220 00:13:30,200 --> 00:13:34,820 downstream of energy, which is the primary source source of 221 00:13:34,910 --> 00:13:37,730 energy for everything here. And then after that, it goes to 222 00:13:37,820 --> 00:13:43,540 compute. So think chips, then chips and data centers. Then 223 00:13:43,900 --> 00:13:47,380 after that, it's the models themselves, harnesses, which is 224 00:13:47,740 --> 00:13:51,850 you know Claude Code or CoWork or Hermes, Open Claw, etc. 225 00:13:52,810 --> 00:13:56,740 Agents that can do the work for you themselves, and then it goes 226 00:13:56,830 --> 00:14:02,040 into identity, building a web of trust, or exactly what 227 00:14:02,130 --> 00:14:05,190 permissions you give any specific agent yourself, 228 00:14:05,820 --> 00:14:10,350 coordinating across them, value transfer and settlement, and so 229 00:14:10,440 --> 00:14:14,400 we can go into each one of those a bit further. But throughout 230 00:14:14,550 --> 00:14:18,360 time, the the entire kind of the supply chain has seen 231 00:14:18,780 --> 00:14:23,630 transaction costs be about 7% of GDP, because there have been 232 00:14:24,200 --> 00:14:28,430 massive dislocations with the amount of available data at each 233 00:14:28,580 --> 00:14:31,970 step of the supply chain, the amount of knowledge that's 234 00:14:31,970 --> 00:14:36,770 available, and I think that right now, all the entire work 235 00:14:37,160 --> 00:14:41,560 chain at this point is really collapsing significantly into 236 00:14:43,060 --> 00:14:47,290 all of the data being extremely available, callable instantly, 237 00:14:47,770 --> 00:14:50,410 and thus a lot less leakage. 238 00:14:50,860 --> 00:14:54,070 We're recording this on the 17th of of 239 00:14:55,120 --> 00:14:58,000 September, and in the last week, there's been a lot of 240 00:14:58,030 --> 00:15:02,820 conversation just around. Is now the time to rein in AI? AI is is 241 00:15:03,240 --> 00:15:06,090 just advancing so quickly, and some of the leaders of the 242 00:15:06,180 --> 00:15:09,750 Frontier Labs are calling for more restraint in the growth. 243 00:15:09,870 --> 00:15:13,470 There have been some incidences that have caused concern about 244 00:15:13,530 --> 00:15:16,560 how fast it's growing and how independent these agents could 245 00:15:16,620 --> 00:15:21,110 become in working against or or just working to achieve a goal 246 00:15:21,500 --> 00:15:26,510 with some disregard of what what constraints may be, how how how 247 00:15:26,600 --> 00:15:28,820 does that factor in? Or I'm sure that you've thought about this 248 00:15:28,940 --> 00:15:32,300 in terms of your your structure and the stack that you're laying 249 00:15:32,480 --> 00:15:35,570 out. What would happen, or what do you see it as even a 250 00:15:35,630 --> 00:15:39,980 possibility that this can that this stack will have a hiccup or 251 00:15:40,040 --> 00:15:44,140 will have a you know a scale back in terms of its evolution 252 00:15:44,170 --> 00:15:44,740 or its growth? 253 00:15:44,920 --> 00:15:48,010 Yeah, I mean there are a lot of people who are who've 254 00:15:48,100 --> 00:15:52,060 been screeching about safety recently, especially Dario with 255 00:15:52,120 --> 00:15:56,890 his "We Must Pace the Frontier" letter recently. I think that 256 00:15:56,920 --> 00:16:00,310 it's mostly marketing for their IPO and saying that they have 257 00:16:00,520 --> 00:16:03,270 something that's so powerful that you know it can't be 258 00:16:03,510 --> 00:16:06,090 released in the wild, and and they're looking for regulatory 259 00:16:06,180 --> 00:16:09,390 capture for having their friends at Meter, who are funded by the 260 00:16:09,420 --> 00:16:13,770 same people, be integrated into their businesses, and thus kind 261 00:16:13,770 --> 00:16:17,730 of add regulatory costs to everything. I think in the the 262 00:16:17,820 --> 00:16:22,640 long term, we're we're all going to have our own AI that's going 263 00:16:22,640 --> 00:16:25,970 to be post trained on our own data and and specific to what we 264 00:16:25,970 --> 00:16:29,690 want to have done. And so with that, there will be an 265 00:16:29,750 --> 00:16:34,310 incredible amount of foundation models out there, and it it 266 00:16:34,400 --> 00:16:38,120 can't necessarily just all go to one person. So with respect to 267 00:16:38,150 --> 00:16:43,810 the actual risks, I think that the risks lie in having a very 268 00:16:43,960 --> 00:16:48,850 significant amount of compute, and thus being able to take 269 00:16:50,350 --> 00:16:53,020 either people who want to achieve things that are bad 270 00:16:53,080 --> 00:16:55,900 because there are bad people in the world, like that will either 271 00:16:56,020 --> 00:17:01,320 try to steal your Bitcoin or you know want to have cyber attacks 272 00:17:01,380 --> 00:17:06,390 on U.S. infrastructure, whether it's countries or just terrorist 273 00:17:06,420 --> 00:17:10,800 groups, and so I think that that is a risk, and I think that you 274 00:17:10,830 --> 00:17:14,250 know that that should be taken somewhat seriously. But I think 275 00:17:14,340 --> 00:17:20,100 that the only way to really have that risk be sufficient is 276 00:17:20,730 --> 00:17:27,260 giving AI a massive amount of compute and allow that to, you 277 00:17:27,290 --> 00:17:31,400 know, essentially go unbounded to attack systems. And so the 278 00:17:31,640 --> 00:17:36,500 only way that that obviously is hap is capable with open source 279 00:17:36,620 --> 00:17:40,990 models. But honestly, I think that there is a relatively low 280 00:17:41,320 --> 00:17:45,070 risk of people running open source models on their their 281 00:17:45,130 --> 00:17:48,220 laptop that are able to like really attack U.S. critical 282 00:17:48,250 --> 00:17:51,370 infrastructure, and so I think that the what it all comes down 283 00:17:51,430 --> 00:17:56,200 to is that these frontier labs are trying to just limit the 284 00:17:56,320 --> 00:18:00,070 amount of capabilities that people can have, so that they 285 00:18:00,100 --> 00:18:02,910 don't need to train their models even more because they have 286 00:18:03,000 --> 00:18:06,240 these upcoming IPOs. I don't. I don't know what you guys think 287 00:18:06,270 --> 00:18:10,290 about that as well. But it feels like a lot of hysteria that 288 00:18:10,410 --> 00:18:13,440 isn't necessarily grounded in anything because we, while we've 289 00:18:13,590 --> 00:18:17,190 seen that one OpenAI hugging face incident, there hasn't been 290 00:18:17,460 --> 00:18:20,010 anything else that's really significant at this point. 291 00:18:20,100 --> 00:18:21,890 I'm actually just shocked at how many times 292 00:18:22,010 --> 00:18:25,730 I'm hearing the the name of the company Hugging Face on on in 293 00:18:25,850 --> 00:18:29,240 traditional media these days, and I've been familiar with and 294 00:18:29,270 --> 00:18:32,030 using the company for about eight or nine years now. I I 295 00:18:32,210 --> 00:18:35,180 used to invest heavily in AI, and so I'm familiar with it. But 296 00:18:35,270 --> 00:18:39,530 yeah, it does feel that there's a you know coordinated effort, 297 00:18:39,590 --> 00:18:44,470 or somebody is just like you know, just piling onto this 298 00:18:44,590 --> 00:18:47,860 news, and it feels unusual. It feels like an unusual time and 299 00:18:47,920 --> 00:18:51,940 space. And I agree with you that I think it would be related to 300 00:18:51,970 --> 00:18:56,200 the IPO or or just the fear that these frontier labs who have 301 00:18:56,260 --> 00:18:58,750 raised so much investment capital and have very 302 00:18:59,680 --> 00:19:02,460 significant valuations that they need to live up to are now 303 00:19:02,910 --> 00:19:07,110 facing pressure from open source labs around the world and 304 00:19:07,230 --> 00:19:10,740 different other models and different other financial models 305 00:19:10,890 --> 00:19:13,050 that could disrupt their business and and make their 306 00:19:13,050 --> 00:19:17,160 valuations tumble down, which is a risk, frankly, to the U.S. 307 00:19:18,240 --> 00:19:21,620 stock market because so much of the market is based on the 308 00:19:21,680 --> 00:19:25,460 valuations of these AI and AI related companies. So, yeah, 309 00:19:25,970 --> 00:19:28,460 interesting. I I would tend to agree. Israel, what about you? 310 00:19:28,580 --> 00:19:28,640 No, 311 00:19:28,790 --> 00:19:31,670 I mean I just don't see how that you can stop something 312 00:19:31,700 --> 00:19:37,580 like this. I mean, even even if some of these claims were true, 313 00:19:37,700 --> 00:19:41,140 which to be clear, and to your point, Liam, I also haven't seen 314 00:19:41,230 --> 00:19:44,560 any hard evidence or data behind this. It's all just kind of been 315 00:19:44,650 --> 00:19:48,520 these stories or yeah claims. Ultimately, so it's hard to 316 00:19:48,550 --> 00:19:52,840 know. But even regardless of that, I mean, I think both 317 00:19:52,990 --> 00:19:55,960 Bitcoin and AI are technologies that are now out there, and you 318 00:19:56,110 --> 00:19:59,470 you can't really stop them. You know, in in a way because of the 319 00:19:59,560 --> 00:20:04,290 game theory. Bitcoin, of course, is a lot more decentralized and 320 00:20:04,380 --> 00:20:08,640 open than than AI. AI is still concentrated in a few global 321 00:20:08,700 --> 00:20:13,200 players, but I mean, ultimately, my mindset around that is you 322 00:20:13,260 --> 00:20:17,910 would need very coordinated global efforts from you know 323 00:20:18,180 --> 00:20:22,280 adversaries, even you know geopolitically, to actually slow 324 00:20:22,370 --> 00:20:27,200 this down and or you know try to pause it, so I just don't see 325 00:20:27,290 --> 00:20:29,330 it. I mean, I think the game theory just doesn't allow it. 326 00:20:29,420 --> 00:20:29,990 Yeah, and I think 327 00:20:30,110 --> 00:20:33,920 that it's almost similar to the crypto wars of the 80s, 328 00:20:34,580 --> 00:20:38,750 back when they were trying to make encryption illegal, weapons 329 00:20:38,840 --> 00:20:43,210 grade, et cetera, and into a lot of the Atomback and other 330 00:20:43,420 --> 00:20:49,150 cypherpunks of that area era really tried to combat that with 331 00:20:49,180 --> 00:20:53,110 their work with PGP. And I think that yeah, I mean there are 332 00:20:53,350 --> 00:20:57,220 people that use encryption for bad things. Like that's I think 333 00:20:57,280 --> 00:21:00,490 that's well known. But I think that just like AI, like 334 00:21:01,330 --> 00:21:04,860 encryption is a tool for privacy, and and it does create 335 00:21:04,950 --> 00:21:08,850 a lot of great things. And similarly, I think that you know 336 00:21:08,970 --> 00:21:12,630 there there will be probably some small negative implications 337 00:21:12,720 --> 00:21:16,110 from AI. But I think that the amount of abundance and goodness 338 00:21:16,230 --> 00:21:19,860 that it will create for everybody out there, from you 339 00:21:19,890 --> 00:21:23,720 know allow causing deflation for goods and services to curing 340 00:21:23,810 --> 00:21:27,320 diseases, allowing people to live healthier lives, saving 341 00:21:27,380 --> 00:21:30,860 them time to work more on what they actually want to work on. 342 00:21:30,950 --> 00:21:34,400 It's going to be a massive net positive, and I don't think that 343 00:21:34,460 --> 00:21:38,540 there is any reasonable way to regulate it at the at this 344 00:21:38,600 --> 00:21:39,230 stage. Lynne 345 00:21:39,320 --> 00:21:42,220 and I are partners at Baselayer Advisors, where 346 00:21:42,310 --> 00:21:44,500 through our experience and network with venture capital 347 00:21:44,560 --> 00:21:47,440 partners, we connect interested investors with unique 348 00:21:47,500 --> 00:21:50,560 opportunities in the space and advise startups on their growth 349 00:21:50,800 --> 00:21:54,250 and fundraising. Visit our website's advisory section to 350 00:21:54,310 --> 00:21:58,090 learn more. Yeah, I largely agree. Well, let's let's maybe 351 00:21:58,300 --> 00:22:02,700 unpack the the different layers to the to to the stack in your 352 00:22:02,760 --> 00:22:05,820 report, Liam, we can just kind of tackle them, you know, two or 353 00:22:05,880 --> 00:22:09,810 three at a time. But you start with electricity and models as 354 00:22:10,200 --> 00:22:13,950 you know as kind of the base, and you hit on some points which 355 00:22:14,400 --> 00:22:16,770 I'd like you to you know expand a little bit on. In the 356 00:22:16,950 --> 00:22:20,390 electricity stack, you get you highlight how we have 357 00:22:20,600 --> 00:22:25,010 bottlenecks. We've underinvested in the infrastructure. You know, 358 00:22:25,100 --> 00:22:28,700 there's a lot of efforts and capital right now in building 359 00:22:29,060 --> 00:22:32,240 out the the sudden burst of demand that we have for for 360 00:22:32,270 --> 00:22:36,590 electricity. Of course, all these data centers that are are 361 00:22:37,130 --> 00:22:40,390 all over the news. How do you view these bottlenecks, and how 362 00:22:40,570 --> 00:22:44,230 does that play into the next layer of the stack, which are 363 00:22:44,290 --> 00:22:47,290 the models themselves, which you argue are only going to get 364 00:22:47,380 --> 00:22:47,620 cheaper? 365 00:22:47,800 --> 00:22:50,470 Yeah, thinking about energy, I think is the biggest 366 00:22:50,530 --> 00:22:55,060 bottleneck because that's heavily regulated industry for 367 00:22:55,120 --> 00:22:59,260 the most part, and so with that, I think that the the grid has 368 00:22:59,590 --> 00:23:02,670 certain restrictions that are going to keep massive amounts 369 00:23:02,730 --> 00:23:07,350 of, you know, power plants, coal mines, etc. that are going to be 370 00:23:07,530 --> 00:23:11,850 held in the queue for a decently long period of time. The grid's 371 00:23:11,880 --> 00:23:14,940 been about flat for the past two decades in terms of the total 372 00:23:15,090 --> 00:23:18,300 energy that it generates, and so I think right now that's the 373 00:23:18,480 --> 00:23:22,070 constraint. Obviously, in the U.S. that's a big focus, and and 374 00:23:22,580 --> 00:23:27,200 more so moving faster in areas like China. But in the end of 375 00:23:27,200 --> 00:23:30,170 the day, there's only so much that they can do, just given its 376 00:23:30,230 --> 00:23:33,080 regulated industry, and that's even why we've seen everything 377 00:23:33,260 --> 00:23:38,870 that's become almost not not quite consensus, but like a lot 378 00:23:38,900 --> 00:23:42,130 of people think that they're going to be orbital data centers 379 00:23:42,250 --> 00:23:46,780 relatively soon because that can get be a way to get out of the 380 00:23:46,960 --> 00:23:50,140 permitting issues that come with respect to everything that's 381 00:23:50,200 --> 00:23:54,280 going on in the U.S. and energy creation, and so I think that 382 00:23:54,460 --> 00:23:57,310 energy is going to remain the bottleneck. But while we've seen 383 00:23:57,460 --> 00:24:01,380 actual supply chain issues with respect to compute itself as it 384 00:24:01,740 --> 00:24:07,530 relates to chips, just because TSMC and folks like them are 385 00:24:07,680 --> 00:24:11,490 pretty much the only folks with enough institutionalized 386 00:24:11,550 --> 00:24:15,810 knowledge to create fabs themselves. The chips still are 387 00:24:15,900 --> 00:24:20,900 getting 50 to 100x better per year, and that's obviously 388 00:24:20,990 --> 00:24:26,060 creating both the ability to issue tokens faster. So 389 00:24:26,780 --> 00:24:30,860 obviously, faster tokens means you can go back and forth with 390 00:24:30,860 --> 00:24:35,570 the AI a lot more, or or just it can work for longer tasks 391 00:24:35,870 --> 00:24:40,070 autonomously, while as just the ability for it to hold more 392 00:24:40,130 --> 00:24:44,710 memory and transfer that at a faster rate, and so energy is 393 00:24:44,890 --> 00:24:48,610 the the biggest bottleneck there. Nvidia and Sarah Bross 394 00:24:48,760 --> 00:24:51,370 and and some other companies are are doing great work on the 395 00:24:51,370 --> 00:24:56,020 compute side, but I think on the energy side, as as it relates to 396 00:24:56,080 --> 00:24:58,570 kind of opportunities that we've been talking about, as it 397 00:24:58,570 --> 00:25:03,600 relates to seeing. Soaking costs go up 25,000% over the past two 398 00:25:03,750 --> 00:25:07,710 years, and trying to invest across the landscape, there 399 00:25:07,950 --> 00:25:10,530 there are really interesting opportunities on the energy 400 00:25:10,650 --> 00:25:14,430 side. I think specifically with respect to storage and 401 00:25:14,910 --> 00:25:18,480 transmission, I think you know base power is an interesting 402 00:25:18,540 --> 00:25:22,130 company doing interesting things at the moment. Compute is 403 00:25:22,250 --> 00:25:25,160 obviously one that's done extremely well, and there will 404 00:25:25,340 --> 00:25:28,910 likely be new disruptions there just because of everything as it 405 00:25:28,910 --> 00:25:32,720 relates to the engineering of of these chips and ultimately more 406 00:25:32,780 --> 00:25:36,050 capacity coming online. I think that both of those are are 407 00:25:36,140 --> 00:25:39,470 investable opportunities, and and we can get into models next. 408 00:25:39,530 --> 00:25:44,470 But I don't necessarily think those are as differentiated, and 409 00:25:45,010 --> 00:25:49,870 you know we'll have a long enough time horizon to return 410 00:25:50,020 --> 00:25:53,290 substantial amounts of cash flow to their actual investors 411 00:25:53,350 --> 00:25:55,780 relative to the nodes of energy and compute. 412 00:25:56,650 --> 00:25:59,140 You know, kind of together with this backlash 413 00:25:59,230 --> 00:26:02,160 or fear about AI and the general public. There's the same thing 414 00:26:02,250 --> 00:26:04,680 about the construction of data centers and forms of 415 00:26:04,710 --> 00:26:08,340 electricity. What would you advise? I mean, it feels like 416 00:26:08,370 --> 00:26:12,090 there's been a slowdown even in traditional energy developed 417 00:26:13,260 --> 00:26:16,560 states like Texas. Abbott's taking kind of a step back on a 418 00:26:16,590 --> 00:26:20,720 data center development. How would you frame it to to help 419 00:26:21,380 --> 00:26:24,170 shift that narrative about how important electricity is, just 420 00:26:24,260 --> 00:26:28,190 overall, and the need for for these foundational systems. 421 00:26:28,430 --> 00:26:31,760 Yeah, I think a lot of people, at least in our circles, 422 00:26:31,790 --> 00:26:35,810 I would say, have seen the chart about how there are no energy 423 00:26:37,250 --> 00:26:41,950 poor and developed countries over time. Like it's directly 424 00:26:42,040 --> 00:26:44,860 correlated with the amount of energy that a country has to the 425 00:26:44,950 --> 00:26:48,730 amount of prosperity and GDP that it has, and so I think that 426 00:26:48,760 --> 00:26:52,600 it's it's obviously crucial. And if you want the your 427 00:26:53,530 --> 00:26:57,730 constituents to be well educated and for them to have better 428 00:26:58,270 --> 00:27:00,960 business opportunities and and better job opportunities 429 00:27:01,020 --> 00:27:03,720 domestically than elsewhere. Then I think that it that is 430 00:27:04,110 --> 00:27:09,360 something that every country municipality should be thinking 431 00:27:09,390 --> 00:27:13,350 about very critically. It's a tough one though because they're 432 00:27:13,560 --> 00:27:16,230 they're incredibly unpopular. I think this is going to be 433 00:27:16,350 --> 00:27:21,710 probably the biggest biggest sticking point in terms of AI at 434 00:27:22,340 --> 00:27:25,490 the next election coming up, because I think something like 435 00:27:25,640 --> 00:27:31,160 80% of people view data centers and AI as unfavorable. I think 436 00:27:31,250 --> 00:27:35,150 that a lot of that has to do, you know, there there are some 437 00:27:35,330 --> 00:27:39,740 noise pollution, but I think that most of them are built far 438 00:27:39,950 --> 00:27:44,290 away from actually where people live, but the real issue is they 439 00:27:44,740 --> 00:27:50,410 see data centers and AI as just one of the the physical 440 00:27:50,650 --> 00:27:56,590 manifestation of the emerging gap between the wages and the 441 00:27:56,650 --> 00:28:00,570 amount of capital that people on the lower shape of the K have 442 00:28:00,720 --> 00:28:06,510 versus the upper shape of the K. They they see you know Elon, Sam 443 00:28:06,630 --> 00:28:12,120 Altman, etc. All these very rich people who you know have a ton 444 00:28:12,150 --> 00:28:16,050 of money, and they just hear on the news that it's because you 445 00:28:16,080 --> 00:28:18,300 know they're stealing from the country, and and they shouldn't 446 00:28:18,330 --> 00:28:20,900 have that, and it should be redistributed. And then they 447 00:28:21,140 --> 00:28:24,290 also hear that AI is going to change their jobs, which I think 448 00:28:24,410 --> 00:28:28,580 is another false narrative. And because of those two are, I 449 00:28:28,700 --> 00:28:31,370 think, the the most popular narratives out there. It's going 450 00:28:31,370 --> 00:28:35,420 to be difficult to ultimately get the people to really change 451 00:28:35,810 --> 00:28:39,740 how they use it, unless or how they view AI, unless they 452 00:28:40,070 --> 00:28:42,940 actually see it in the benefits themselves, which they they 453 00:28:43,060 --> 00:28:46,420 largely haven't yet, outside of just having a quicker, faster 454 00:28:46,570 --> 00:28:51,490 Google. And ultimately, the the politicians probably want to say 455 00:28:51,670 --> 00:28:55,510 whatever the people actually want to get elected. And so, I 456 00:28:55,660 --> 00:28:57,730 think it's going to be a challenging topic over over 457 00:28:57,820 --> 00:29:00,870 time. And I don't necessarily have this silver bullet, whether 458 00:29:00,930 --> 00:29:05,190 it's narrative on how to shape this better, what what do you 459 00:29:05,280 --> 00:29:05,820 guys think? 460 00:29:06,180 --> 00:29:08,520 I'm not sure either on where this goes, Liam. All I 461 00:29:08,670 --> 00:29:13,020 know is that the negative narrative around it will 462 00:29:13,140 --> 00:29:18,120 continue. So I think that that's just not going away. You know, 463 00:29:18,210 --> 00:29:24,740 maybe it adapts. We saw Bitcoin Bitcoin mining infrastructure 464 00:29:24,800 --> 00:29:28,190 adapt, you know, in the last 10 years, and plugged into more, 465 00:29:28,880 --> 00:29:32,690 you know, rural energy sources, and it's gotten a little bit 466 00:29:32,750 --> 00:29:38,150 more renewable energy focused. So it's, you know, I think 467 00:29:38,240 --> 00:29:41,890 ultimately it's kind of one of those things where the market 468 00:29:42,010 --> 00:29:47,050 will work it out as as it as it should, you know. But I do think 469 00:29:47,350 --> 00:29:50,830 the regulatory headwinds are not going away anytime soon. 470 00:29:51,820 --> 00:29:53,590 Yeah, I agree with Israel. I think that 471 00:29:53,710 --> 00:29:56,890 Bitcoin mining has given a great example of how a narrative can 472 00:29:57,010 --> 00:30:00,390 shift relatively quickly from some. Something that was 473 00:30:00,480 --> 00:30:03,540 misunderstood and misinterpreted to something that now is you 474 00:30:03,600 --> 00:30:07,050 know we understand that it can actually improve the pace of 475 00:30:07,140 --> 00:30:10,950 renewable development in electricity. I think I'm hoping 476 00:30:11,070 --> 00:30:15,000 that after the midterm U.S. elections that the narrative 477 00:30:15,030 --> 00:30:17,310 will shift a little bit. That'll it'll take the pressure off at 478 00:30:17,340 --> 00:30:20,930 least for a year and a half, and then. But I'm also wondering if 479 00:30:20,960 --> 00:30:24,320 this just gives opportunities. I mean, our podcast tends to look 480 00:30:24,500 --> 00:30:27,470 at these developments on more of a global scale than just U.S. 481 00:30:27,530 --> 00:30:30,650 centric. So, you know, perhaps there are opportunities in other 482 00:30:30,740 --> 00:30:34,520 regions and other areas to help do some of the AI build out that 483 00:30:34,580 --> 00:30:37,910 would increase the opportunities in those emerging economies 484 00:30:38,180 --> 00:30:41,890 where they do need more electricity. They need more to 485 00:30:42,130 --> 00:30:45,370 to improve the prosperity and the GDP. That this could be a 486 00:30:45,460 --> 00:30:47,590 source of opportunity for other regions. 487 00:30:48,100 --> 00:30:50,590 Yeah, I think that's fantastic point. Very well said, 488 00:30:50,800 --> 00:30:55,750 and something that you know as there are differing opinions, 489 00:30:55,900 --> 00:30:59,740 whether it's the U.S. versus China, and how those are so 490 00:30:59,860 --> 00:31:03,570 diametrically opposed with the public perception of AI. I think 491 00:31:03,690 --> 00:31:07,170 that outside of the U.S. there are a ton of countries. I think 492 00:31:07,260 --> 00:31:10,350 South Korea just actually said that they're going to allow for 493 00:31:10,860 --> 00:31:14,610 something like $8 billion worth of free AI for their people too. 494 00:31:15,180 --> 00:31:17,790 I think that that's a an incredibly smart strategy in 495 00:31:17,850 --> 00:31:22,760 order to you know have your you know citizens be able to retrain 496 00:31:22,820 --> 00:31:26,030 themselves in order to bring your country further ahead. I 497 00:31:26,090 --> 00:31:28,880 think that they should be teaching how to use these types 498 00:31:28,970 --> 00:31:32,960 of tools in schools, and and that will help improve 499 00:31:33,050 --> 00:31:36,380 perception, and then also provide massive economic 500 00:31:36,530 --> 00:31:37,670 opportunity, as you mentioned. 501 00:31:38,630 --> 00:31:41,230 Yeah, and and I mean on that point worth highlighting as 502 00:31:41,410 --> 00:31:44,230 well, and I'm not too familiar with the numbers, but I did see 503 00:31:44,440 --> 00:31:48,610 recently El Salvador that they've been implementing AI at 504 00:31:48,670 --> 00:31:52,630 the educational level at a very early stage, I believe. And 505 00:31:53,710 --> 00:31:56,290 again, I'm not familiar with the stats, but I think they ranked 506 00:31:57,010 --> 00:32:00,690 pretty high in recent testing of their young kids. So it's having 507 00:32:01,290 --> 00:32:04,680 clearly big benefits to those who are leaning into this 508 00:32:04,770 --> 00:32:08,250 technology. I mean, ultimately, the U.S. is the powerhouse 509 00:32:08,490 --> 00:32:12,030 economically that it is today because it, for the most part, 510 00:32:12,120 --> 00:32:15,180 embraced the internet early on, and so as a consequence, you 511 00:32:15,180 --> 00:32:18,360 know, now you have some of the largest companies and all you 512 00:32:18,360 --> 00:32:21,680 know multiple trillion dollar companies that are, you know, 513 00:32:21,770 --> 00:32:25,610 that were built from the Internet Protocol, and the U.S. 514 00:32:25,670 --> 00:32:28,460 is has of course reaped massive benefits from that. So I think, 515 00:32:29,270 --> 00:32:31,490 you know, those those who lean into this will certainly 516 00:32:31,550 --> 00:32:36,440 benefit. But well, maybe getting back to the to the stack itself, 517 00:32:36,560 --> 00:32:39,140 Liam. So we went through the electricity component, the 518 00:32:39,650 --> 00:32:43,270 models themselves, as I understand it, you know. Correct 519 00:32:43,300 --> 00:32:45,820 me if I'm wrong, but you you view these as a little more 520 00:32:45,940 --> 00:32:50,110 fluid. They'll kind of you know they'll come and go likely for 521 00:32:50,170 --> 00:32:53,020 the foreseeable future, both closed source and open source. 522 00:32:53,650 --> 00:32:57,970 Then you get into a layer of the stack called the harness, which 523 00:32:58,960 --> 00:33:01,740 is, in other words, perhaps you know the infrastructure layer. 524 00:33:01,950 --> 00:33:05,970 Let's say the tools and and what you actually need to use some of 525 00:33:06,030 --> 00:33:09,420 these models. And you've you view this as one of the most 526 00:33:09,900 --> 00:33:13,620 maybe sticky or investable layers. Again, correct me if I'm 527 00:33:13,680 --> 00:33:17,130 wrong on on any of this. And then we you know you kind of get 528 00:33:17,190 --> 00:33:20,630 into how the agents interact. You know on top of the models 529 00:33:20,750 --> 00:33:24,290 and and this harness layer, is that more or less the thesis, or 530 00:33:24,500 --> 00:33:25,850 what? What can you add to that? 531 00:33:25,970 --> 00:33:28,850 Yeah, for sure. There, with respect to the models 532 00:33:28,910 --> 00:33:32,150 themselves, they're obviously those have raised the most 533 00:33:32,180 --> 00:33:35,810 amount of capital to date, and I think that they're they get a 534 00:33:35,840 --> 00:33:40,280 lot of noise because they that's kind of what the back end is 535 00:33:40,340 --> 00:33:44,500 that powers all of this, and some of them have also like 536 00:33:44,590 --> 00:33:51,040 obviously cloud code and and codecs are harnesses for the 537 00:33:51,040 --> 00:33:55,780 anthropics and open AIs of the world who offer just a great 538 00:33:56,140 --> 00:34:00,130 user experience within their tools. But ultimately, I think 539 00:34:00,220 --> 00:34:04,230 that a lot of the the models are a little bit unsustainable 540 00:34:04,350 --> 00:34:07,800 because of the economics that go into it, and what I mean by that 541 00:34:07,800 --> 00:34:12,180 is, if you're paying for a $200 cloud subscription right now, it 542 00:34:12,180 --> 00:34:15,270 actually costs them. If you're using it into the fullest 543 00:34:15,360 --> 00:34:20,190 extent, it's costing them about 8000 to $9,000 a month, and so 544 00:34:20,400 --> 00:34:23,390 that's a massively negative margin. But why do they do that? 545 00:34:23,450 --> 00:34:28,700 It's because they are taking all of the data about how consumers 546 00:34:29,000 --> 00:34:32,990 and businesses use their information, and they will then 547 00:34:33,350 --> 00:34:37,070 have all of that data that's readily available to train their 548 00:34:37,130 --> 00:34:41,380 next model or use it in order to go out into a different 549 00:34:41,440 --> 00:34:45,610 horizontal verticals, and so like folks like Figma have come 550 00:34:45,790 --> 00:34:49,300 on and been partners with them, and use Claude as a backend, and 551 00:34:49,330 --> 00:34:54,280 then Claude goes and does Claude design, and so I think with that 552 00:34:54,340 --> 00:34:58,090 it's it's interesting to do because they have the API that 553 00:34:58,150 --> 00:35:02,430 they can train off of that. Has like 90% margins, but they lose 554 00:35:02,490 --> 00:35:08,430 a lot of the the actual money on the subscription, and then at 555 00:35:08,460 --> 00:35:11,580 the same time they have to put hundreds of millions of dollars 556 00:35:11,790 --> 00:35:17,310 into training runs, and then that is extremely like 557 00:35:17,460 --> 00:35:21,680 unsustainable. If there can be other models out there that can 558 00:35:21,740 --> 00:35:25,790 just take all of the information that you get from a training run 559 00:35:26,210 --> 00:35:30,230 and essentially use that to train their models themselves 560 00:35:30,350 --> 00:35:34,610 too, or a 10th or you know 1/20 of the cost of doing it 561 00:35:34,610 --> 00:35:37,850 themselves. And so it's it's widely rumored right now that 562 00:35:37,850 --> 00:35:40,610 many of those models are actually just not releasing 563 00:35:40,760 --> 00:35:44,290 their latest ones to the public, and folks like Anthropic are 564 00:35:44,350 --> 00:35:48,070 actually going more into drug discovery. And so, I think that 565 00:35:48,250 --> 00:35:52,450 while consumer and business serving of you know creating 566 00:35:52,540 --> 00:35:55,210 models and offering it to the public and having them pay for 567 00:35:55,300 --> 00:35:59,200 it is something that's not necessarily going to be a great 568 00:35:59,260 --> 00:36:01,890 long term business model because of the cost of distillation, 569 00:36:01,980 --> 00:36:05,370 don't are significantly lower than the cost of training models 570 00:36:05,550 --> 00:36:09,930 yourself. There, that could be a tough business model. So I think 571 00:36:10,380 --> 00:36:13,770 I wouldn't be super excited about owning Anthropic or or 572 00:36:13,920 --> 00:36:18,630 OpenAI for that type of consumer-facing business. The 573 00:36:18,780 --> 00:36:21,860 harnesses and agents are what's actually the most valuable part? 574 00:36:22,460 --> 00:36:26,720 It's the if you've used Hermes or Open Claw or or Open Code, 575 00:36:28,160 --> 00:36:32,360 they are the interface that you use for any sort of model in the 576 00:36:32,360 --> 00:36:37,070 backend, either one that you host yourself or an API that can 577 00:36:37,130 --> 00:36:40,870 be switched between them, and that's really valuable because 578 00:36:41,200 --> 00:36:44,680 it actually is the interface, and it can significantly 579 00:36:44,830 --> 00:36:48,640 decrease the amount of tokens or increase if you have the wrong 580 00:36:48,760 --> 00:36:52,120 harness and the wrong tools that are associated with it. Whether 581 00:36:52,210 --> 00:36:59,020 it's connecting it to your Excel, PowerPoint, email, all of 582 00:36:59,020 --> 00:37:04,560 the other tools that you likely have is is an average user, it's 583 00:37:04,770 --> 00:37:08,640 able to call them and and get hooked up and do the essentially 584 00:37:08,880 --> 00:37:13,110 operating system and interface for you to work with essentially 585 00:37:13,560 --> 00:37:16,170 all of your intelligence, and that's where you can actually 586 00:37:16,260 --> 00:37:19,350 spawn agents and have them be focused on different tasks like 587 00:37:19,860 --> 00:37:24,050 design versus writing copy versus financial modeling and 588 00:37:24,080 --> 00:37:28,160 Excel, and I think that's where a lot of the durable value 589 00:37:28,280 --> 00:37:31,490 lives. Although many of those are still being given away for 590 00:37:31,610 --> 00:37:35,330 free, just because the open source community is so strong 591 00:37:35,450 --> 00:37:38,660 and and they want to really look for bugs, whether it's Open Claw 592 00:37:38,750 --> 00:37:43,090 or Hermes or or whatever it is out there, there are additional 593 00:37:43,150 --> 00:37:47,560 hosting costs that you can benefit from for for those types 594 00:37:47,650 --> 00:37:53,050 of harnesses. But I think that those are the kind of main areas 595 00:37:53,170 --> 00:37:55,840 that that I'm really interested in on my side. 596 00:37:55,930 --> 00:37:58,180 I think that's a great way to kind of move on to 597 00:37:58,210 --> 00:38:01,830 the next level of the stack, which I'm going to put these 598 00:38:01,920 --> 00:38:04,560 together the way Israel kind of combines some of the 599 00:38:04,560 --> 00:38:08,160 infrastructure part of it, but you know the the next layer of 600 00:38:08,220 --> 00:38:10,830 the stack in your report are the agents. And I first of all I 601 00:38:10,890 --> 00:38:13,290 just want to say we'll include a link to the report and to the 602 00:38:13,320 --> 00:38:16,740 website in here. I'd encourage everyone to really take a deep 603 00:38:16,800 --> 00:38:20,100 dive into this because it's fascinating. But to the next the 604 00:38:20,310 --> 00:38:23,180 next layer would be the agents themselves, and they're the the 605 00:38:25,220 --> 00:38:29,870 artificial intelligence that has a goal or has a task or 606 00:38:29,900 --> 00:38:32,960 repetitive task that is out there actually creating the 607 00:38:32,990 --> 00:38:35,630 productivity of it. And associated with that is 608 00:38:35,630 --> 00:38:40,040 identity. So then there becomes the question of, you know, is an 609 00:38:40,220 --> 00:38:45,040 individual AI or an agent has have has a unique identity, or 610 00:38:45,100 --> 00:38:48,190 how do you how do you know which agent is doing which work? And 611 00:38:48,220 --> 00:38:50,770 then on top of that is the coordination among agents. So, 612 00:38:51,370 --> 00:38:54,790 can you can you kind of take a look at that those three layers 613 00:38:54,850 --> 00:38:57,520 of the stack, how they interact and how they evolve one from the 614 00:38:57,640 --> 00:39:01,620 other, from agent to identity to coordination, and is this kind 615 00:39:01,650 --> 00:39:04,290 of the the fear factor that we're seeing right now, or how 616 00:39:04,350 --> 00:39:07,290 do we? I mean, but also the productivity. Like, how much has 617 00:39:07,380 --> 00:39:11,190 that is that unleashing productivity today, and what is 618 00:39:11,280 --> 00:39:13,770 the potential for that going forward? Because I really see 619 00:39:13,890 --> 00:39:17,610 this as kind of the the the the kind of goldmine and how 620 00:39:18,390 --> 00:39:21,200 businesses can unlock potential and individuals can unlock 621 00:39:21,260 --> 00:39:24,860 potential by by utilizing these this agent stack. 622 00:39:26,299 --> 00:39:30,199 Yeah, for sure. I think that right now agents it depends 623 00:39:30,259 --> 00:39:32,749 on how you interact with them, but a lot of them are just 624 00:39:32,809 --> 00:39:37,039 internal. You speak with them on your computer, and they go do a 625 00:39:37,129 --> 00:39:40,449 task. They start to get really interesting when you embed them 626 00:39:40,569 --> 00:39:44,259 with either your company workflows and the agents that 627 00:39:44,379 --> 00:39:48,099 you're interacting with are interacting with other agents 628 00:39:48,129 --> 00:39:51,279 that you're that are created by other folks at your company, or 629 00:39:51,639 --> 00:39:55,749 just interacting with any other agent on the internet for a 630 00:39:56,469 --> 00:40:01,769 shared group of tasks. And so, one of the. Ones was was lobster 631 00:40:02,039 --> 00:40:05,549 claw or something like that about them all getting 632 00:40:05,639 --> 00:40:08,729 identities and and posting on their own Facebook or or 633 00:40:08,819 --> 00:40:13,469 whatever but I think that a lot more of it will be in the future 634 00:40:14,069 --> 00:40:20,159 folks who host a certain amount of compute and they can offer 635 00:40:20,459 --> 00:40:24,709 their agents and to go out and do things in the real world, in 636 00:40:24,829 --> 00:40:27,319 order, and that could be anything from finding 637 00:40:27,529 --> 00:40:31,129 differentiated research that isn't just available and and 638 00:40:31,219 --> 00:40:34,789 paying for that through micro payments, or you know whether 639 00:40:34,999 --> 00:40:40,579 it's being a world class copywriter for advertising and 640 00:40:41,509 --> 00:40:44,229 getting hired by other folks out there because they have a 641 00:40:44,619 --> 00:40:48,969 reputation for doing that at a very high degree. I think that 642 00:40:48,969 --> 00:40:51,669 that's where it starts to get interesting as it relates to 643 00:40:51,909 --> 00:40:56,049 identity and having things like like Nostr on the back end that 644 00:40:56,109 --> 00:41:01,469 will associate end pubs with different agents out there, and 645 00:41:01,589 --> 00:41:05,429 giving them a list, like essentially webs of trust, to be 646 00:41:05,579 --> 00:41:07,769 able to coordinate with different agents. Whether it's 647 00:41:07,949 --> 00:41:12,719 something like Nostr or other protocols out there, like to 648 00:41:12,779 --> 00:41:13,049 date. 649 00:41:13,199 --> 00:41:15,779 Excuse me, Liam. Can you just explain to our 650 00:41:15,839 --> 00:41:18,209 audience that may not be familiar with Nostr what that 651 00:41:18,509 --> 00:41:18,599 is? 652 00:41:18,630 --> 00:41:21,740 Yeah, for sure. It's a it's a coordination system for 653 00:41:24,050 --> 00:41:26,660 there. There's essentially public and private keys, similar 654 00:41:26,780 --> 00:41:29,810 to what Bitcoin has. You know, your private key is your 655 00:41:30,320 --> 00:41:35,210 address. You're proving that you either have the Bitcoin or on 656 00:41:35,240 --> 00:41:39,290 the Nostra side, proving that you own the identity, and then 657 00:41:39,380 --> 00:41:44,110 the public address is essentially your name or a 658 00:41:44,200 --> 00:41:47,020 string of characters that can be associated with the name, and 659 00:41:47,050 --> 00:41:50,260 that can be the agent. So it's tied to the activity that you 660 00:41:50,320 --> 00:41:53,620 do. Or on the Bitcoin side, it's an address that you can use to 661 00:41:54,460 --> 00:42:00,430 receive Bitcoin. And so I think that similarly, like tools like 662 00:42:00,580 --> 00:42:05,550 Jack at Square have created with Buzz, it's allowed for different 663 00:42:05,940 --> 00:42:10,680 agents to interact with your entire team, and so that uses 664 00:42:10,800 --> 00:42:15,330 Nostra as the backend, and it can include different folks who 665 00:42:15,630 --> 00:42:20,400 join your channel in Buzz, and one can be focused on marketing, 666 00:42:20,550 --> 00:42:23,480 one could be focused on sales and go-to-market, or 667 00:42:24,860 --> 00:42:28,880 engineering, product, security, and they can all have different 668 00:42:29,000 --> 00:42:31,700 identities and be different, attached with different things 669 00:42:31,790 --> 00:42:35,270 over time. Similarly, in the real world, whether it's 670 00:42:35,900 --> 00:42:40,720 actually you know paying for different security reviews by 671 00:42:40,840 --> 00:42:44,530 very capable agents who have built up a corpus of knowledge 672 00:42:44,650 --> 00:42:48,190 over time and have different efficiencies as it relates to 673 00:42:48,280 --> 00:42:52,570 skills and and knowledge. I think that similarly, there will 674 00:42:52,600 --> 00:42:58,030 be a lot of value transferred by taking an identity, building up 675 00:42:58,270 --> 00:43:03,510 a history of work that it's done over time, and then using that 676 00:43:03,570 --> 00:43:07,740 to coordinate what you should, what you're willing to pay for 677 00:43:07,860 --> 00:43:09,120 for different future tasks. 678 00:43:09,269 --> 00:43:13,139 Yeah, and the productivity unlock is is going 679 00:43:13,199 --> 00:43:17,399 to be. It's it's still something I have a hard time grasping 680 00:43:17,849 --> 00:43:20,639 because I'm just seeing it in real time, and and it's it's 681 00:43:20,669 --> 00:43:24,259 just absolutely exponential. I mean, a bit of a side here. I 682 00:43:24,319 --> 00:43:28,279 have a I have a packaging business for that that I work on 683 00:43:28,339 --> 00:43:31,789 for restaurants, small coffee shops, those kind of things. And 684 00:43:31,909 --> 00:43:35,359 we're just starting to play around with some of these agent 685 00:43:35,389 --> 00:43:39,259 tools. So specifically Grokbot, and there's a few right. And to 686 00:43:39,439 --> 00:43:43,719 your point of these identities, I mean, you you start to see the 687 00:43:43,779 --> 00:43:47,889 the the real impact of. So we have you know a sales agent, we 688 00:43:48,039 --> 00:43:51,729 have a design agent, we have an accounting agent that's starting 689 00:43:51,819 --> 00:43:56,169 to do a lot of this backend internal work for us. I mean, a 690 00:43:56,229 --> 00:43:58,809 lot of people make parallels to just thinking of them as 691 00:43:58,929 --> 00:44:04,949 coworkers, and in practice, it's true, right? So you you kind of 692 00:44:04,949 --> 00:44:08,249 extrapolate that to the actual changes in the economy and 693 00:44:08,279 --> 00:44:10,979 productivity gains, and and it's it's absolutely mind blowing. 694 00:44:11,039 --> 00:44:13,199 But anyways, I don't mean to deviate. 695 00:44:13,290 --> 00:44:16,260 I was just curious: have you guys seen the Hugging Face? 696 00:44:16,410 --> 00:44:19,590 Just speaking because we came out with it earlier, the the 697 00:44:19,860 --> 00:44:23,540 happy robot thing, the the little ducks that came out last 698 00:44:23,570 --> 00:44:23,750 week. 699 00:44:24,560 --> 00:44:25,250 I have not, Lynne. 700 00:44:25,400 --> 00:44:25,790 Have not. 701 00:44:26,899 --> 00:44:30,289 Okay. Well, it's it's actually funny because you know, 702 00:44:31,159 --> 00:44:33,979 Hugging Face, not like a Bitcoin company or anything, but they 703 00:44:34,069 --> 00:44:37,909 came out with Nostr as a protocol. They they created 704 00:44:37,939 --> 00:44:41,799 these essentially robots that are shaped like docks that are 705 00:44:42,489 --> 00:44:45,099 entirely open source, and you can put your own models in them 706 00:44:45,549 --> 00:44:48,579 for extremely low cost, and they will recursively learn over 707 00:44:48,669 --> 00:44:53,259 time. And they use Nostra as a backend to coordinate the 708 00:44:53,469 --> 00:44:56,289 identity of all the different robots, and then interact with 709 00:44:56,829 --> 00:44:59,949 each other in the real world. And so, I think that that. 710 00:45:00,840 --> 00:45:03,300 It's it's obviously very early, but it's just a sign of 711 00:45:03,390 --> 00:45:06,660 what's to come as it relates to not just you know working with 712 00:45:06,810 --> 00:45:10,920 agents on the computer, but how it actually identifies ties to 713 00:45:11,160 --> 00:45:14,430 actual robots, whether it relates to you know them having 714 00:45:14,940 --> 00:45:17,940 being in Amazon facilities or actually out in the real world 715 00:45:18,090 --> 00:45:21,620 in order to understand where they are with certain tasks and 716 00:45:22,010 --> 00:45:25,400 being able to coordinate exactly who they are over time, too. I 717 00:45:25,460 --> 00:45:28,850 think that it's it's a little bit early, but it's fascinating 718 00:45:29,060 --> 00:45:32,090 to if you extrapolate that out a few years, what what it really 719 00:45:32,180 --> 00:45:35,600 means. Yeah, to your point, I mean this identity and and 720 00:45:35,690 --> 00:45:38,960 coordination component. I mean once that that just takes us 721 00:45:39,020 --> 00:45:41,680 into the actual agent to agent economy, right? Which we're 722 00:45:41,860 --> 00:45:45,580 still at the very, very, very early stages of. And once that 723 00:45:45,670 --> 00:45:48,280 starts to take off, then I mean, yeah, we we move from it just 724 00:45:48,280 --> 00:45:52,810 being, I guess, this internal tool that works for us as humans 725 00:45:52,900 --> 00:45:57,040 and coordinates our efforts to to really handing off complete 726 00:45:57,130 --> 00:46:01,590 tasks and and agents kind of you know really having this value 727 00:46:01,650 --> 00:46:05,490 transfer and settlement and and trading going on between them, 728 00:46:05,610 --> 00:46:10,710 which is just a whole another layer to it. But Lynne, sorry, I 729 00:46:10,770 --> 00:46:12,900 think you were going to add something as well. 730 00:46:12,990 --> 00:46:15,660 This is something that, frankly, Israel 731 00:46:15,660 --> 00:46:17,610 and I have been talking about for quite a while. Is just 732 00:46:17,700 --> 00:46:22,370 opportunities with Bitcoin and and the fact that maybe the 733 00:46:22,490 --> 00:46:26,240 agentic economy and agents themselves could be that unlock 734 00:46:26,390 --> 00:46:29,150 for medium of exchange that maybe we've been looking at the 735 00:46:29,270 --> 00:46:33,740 wrong place instead of having people in evolved countries all 736 00:46:33,740 --> 00:46:36,860 of a sudden switch their payment system to Bitcoin maybe it's not 737 00:46:37,040 --> 00:46:41,560 people maybe it's agents of if you can dive deeper into the 738 00:46:41,650 --> 00:46:44,860 whole value transfer layer of your stack and settlements and 739 00:46:44,890 --> 00:46:45,700 how that ties in, 740 00:46:45,850 --> 00:46:48,190 yeah, for sure. I think that taking a step back too, 741 00:46:48,430 --> 00:46:52,300 where there are two aspects to this too. It's both as this 742 00:46:52,510 --> 00:46:56,140 really unfolds, are you going to save your money because there's 743 00:46:56,170 --> 00:46:59,740 going to be so much deflation out there as it relates to just 744 00:46:59,740 --> 00:47:02,610 the amount of productivity that's coming out as it relates 745 00:47:02,700 --> 00:47:05,040 to every part of the stack getting better and better over 746 00:47:05,100 --> 00:47:09,570 time, and so it really is important to look hard at scarce 747 00:47:09,720 --> 00:47:13,680 assets, whether it's gold or Bitcoin, which is probably my my 748 00:47:13,800 --> 00:47:16,020 personal favorite. But I think that both will do well over 749 00:47:16,110 --> 00:47:21,440 time, and and if you aren't really allocating to scarce 750 00:47:21,650 --> 00:47:24,230 assets, then you're going to see a lot of the gains inflated 751 00:47:24,260 --> 00:47:28,970 away. When looking back at the initial industrial revolution, 752 00:47:29,660 --> 00:47:32,720 while there was so much created and with respect to all the 753 00:47:32,750 --> 00:47:36,440 railroads and a lot of innovations with electricity and 754 00:47:37,670 --> 00:47:40,450 steel, one of the best investments was just to hold 755 00:47:40,540 --> 00:47:43,660 money because it was backed by gold, and there was actually 756 00:47:43,810 --> 00:47:46,660 real deflation there, and so you saw a lot of your purchasing 757 00:47:46,720 --> 00:47:49,720 power increase over time, and didn't have to necessarily pick 758 00:47:49,930 --> 00:47:52,960 what was going to be the best performing asset versus like one 759 00:47:53,050 --> 00:47:57,760 steel company, railroad company, etc. And so that's that's one 760 00:47:57,790 --> 00:48:02,730 piece, but the other one is in a world where you know everything 761 00:48:02,760 --> 00:48:07,380 is kind of constrained by energy and compute. It's almost going 762 00:48:07,410 --> 00:48:12,780 to move towards electricity and how you meter that in terms of 763 00:48:13,170 --> 00:48:20,190 paying for it over time. And so there will be essentially value 764 00:48:20,280 --> 00:48:25,220 tasks that are so small but so critical that will be too small 765 00:48:25,340 --> 00:48:29,480 for credit cards to pay, and so during that it will be crucial 766 00:48:29,570 --> 00:48:34,280 to have internet-native money and money that's native to the 767 00:48:34,430 --> 00:48:38,720 internet. Stablecoins, I think, will play a role there to start, 768 00:48:38,810 --> 00:48:41,890 just because some people are familiar with the dollar and the 769 00:48:42,010 --> 00:48:45,010 network effects that it has, especially offboarding and 770 00:48:45,100 --> 00:48:49,300 onboarding in the U.S. to to different banks. But the world 771 00:48:49,360 --> 00:48:53,530 isn't just U.S. centric. It needs money that's neutral. It 772 00:48:53,950 --> 00:48:57,010 needs money that will actually keep its value over time. And so 773 00:48:57,100 --> 00:49:00,610 I think that that will mostly move towards Bitcoin over time. 774 00:49:01,570 --> 00:49:05,670 There are obviously challenges with really bootstrapping it 775 00:49:05,730 --> 00:49:09,870 right now, and because not not many people unfortunately use it 776 00:49:09,930 --> 00:49:14,040 as payment on a day to day basis, and especially not with 777 00:49:14,190 --> 00:49:18,750 instant settlement. Obviously, Bitcoin takes whatever 10 ish 778 00:49:18,930 --> 00:49:23,090 minutes or so, which may or may not be enough for certain 779 00:49:23,180 --> 00:49:26,930 activities as as AI really accelerates, but I think that 780 00:49:26,960 --> 00:49:30,800 there are a bunch of innovative L 2s out there, and and it's 781 00:49:30,860 --> 00:49:33,530 still honestly pretty early days for those. But whether it's 782 00:49:33,830 --> 00:49:37,970 Lightning, Arc, Spark, and the connectivity between all of 783 00:49:38,060 --> 00:49:40,870 them, I think that that's ultimately the direction that 784 00:49:41,020 --> 00:49:44,830 we're going to move with respect to using micro payments over 785 00:49:44,890 --> 00:49:48,850 time, both because it's it can't be seized, and you know most 786 00:49:48,910 --> 00:49:51,190 people just don't live in the U.S. and don't really have a lot 787 00:49:51,220 --> 00:49:52,900 of use for dollars on a day-to-day basis. 788 00:49:53,199 --> 00:49:56,349 I want to take a quick pause here and and just frame 789 00:49:56,469 --> 00:50:01,409 something because I notice and my suspicion is. A lot of people 790 00:50:02,039 --> 00:50:05,249 sometimes can get confused because we hear inflation and 791 00:50:05,309 --> 00:50:08,639 deflation used in the same, you know, for example, research 792 00:50:08,789 --> 00:50:13,289 papers or arguments, and it can be a bit confusing, right? And 793 00:50:13,409 --> 00:50:17,099 and it's mostly due to the fact that these are kind of competing 794 00:50:17,189 --> 00:50:21,139 forces that we we have. You have deflation as it relates to 795 00:50:21,199 --> 00:50:25,639 technology and inflation as it relates to money. Can you can 796 00:50:25,669 --> 00:50:28,939 you help frame that for for someone who's listening and and 797 00:50:28,999 --> 00:50:32,149 might sometimes get a little bit confused between the words 798 00:50:32,209 --> 00:50:33,259 inflation and deflation? 799 00:50:33,410 --> 00:50:37,520 Yeah, for sure. Whenever I hear about that, I think about 800 00:50:38,030 --> 00:50:40,810 have you guys seen the chart from like the change in prices 801 00:50:40,870 --> 00:50:46,030 since the year 2000, and it's like TVs, iPhones, etc. all 802 00:50:46,330 --> 00:50:50,590 deflating really fast. And then it's on the top side, above 803 00:50:50,680 --> 00:50:54,850 average CPI. It's like healthcare services, education, 804 00:50:55,330 --> 00:51:00,630 etc. And a lot of that is, you know, the the bottom half of the 805 00:51:00,870 --> 00:51:06,540 curve is due to innovation technologically, which can bring 806 00:51:06,600 --> 00:51:10,620 down the cost of the cost of products and services, and and 807 00:51:10,620 --> 00:51:13,980 that's why TVs, etc. get cheaper over time, and that's the 808 00:51:14,130 --> 00:51:17,190 natural state of the economy. People will always get more 809 00:51:17,250 --> 00:51:21,020 productive; otherwise, they're going to get competed out of 810 00:51:21,080 --> 00:51:26,870 their business and lose their job, etc. The so deflation is 811 00:51:26,930 --> 00:51:30,860 the natural state, but there is monetary inflation that is 812 00:51:31,010 --> 00:51:34,460 increasing anywhere from I think it's average at 7% a year over 813 00:51:34,580 --> 00:51:38,630 the past 100 years, and going to likely accelerate further from 814 00:51:38,690 --> 00:51:43,060 here as our debt and deficits start to get to almost untenable 815 00:51:43,960 --> 00:51:49,000 levels, and so that's monetary inflation is is a is a tough 816 00:51:49,180 --> 00:51:53,140 thing for many people because wages and everything that you 817 00:51:53,320 --> 00:51:55,870 own in terms of asset prices doesn't necessarily keep up as 818 00:51:55,930 --> 00:52:00,730 fast. And then on top of that, we see industries that have you 819 00:52:00,760 --> 00:52:04,530 know you can say highly regulated is a friendly way, and 820 00:52:05,040 --> 00:52:09,000 have regulatory capture as a the more pessimistic view. But 821 00:52:09,000 --> 00:52:12,210 everything from healthcare services, education, where 822 00:52:12,270 --> 00:52:15,180 there's essentially government monopolies on how they work, 823 00:52:15,630 --> 00:52:19,230 those tend to see a lot less innovation and thus deflation in 824 00:52:19,500 --> 00:52:23,510 prices, or they are just tied to incentives where they actually 825 00:52:23,600 --> 00:52:26,120 increase prices over time because that's the only way for 826 00:52:26,180 --> 00:52:30,320 those companies to make more profit. And so I think that as 827 00:52:30,500 --> 00:52:33,200 as you look out into the future and try to extrapolate a little 828 00:52:33,230 --> 00:52:38,540 bit more, there will continue to be a lot of technological 829 00:52:38,630 --> 00:52:43,390 deflation as it relates to just industries that are closer to 830 00:52:43,660 --> 00:52:47,980 completely free and have less regulation on them, whereas 831 00:52:48,130 --> 00:52:51,820 those that are heavily regulated will probably be a little bit 832 00:52:51,880 --> 00:52:56,020 closer to the actual monetary inflation rate, which which sees 833 00:52:56,200 --> 00:52:57,820 purchasing power decline over time. 834 00:52:57,940 --> 00:53:00,750 When deflation hits, not just the core 835 00:53:00,960 --> 00:53:04,710 technologies like a television or a component, but hits 836 00:53:04,920 --> 00:53:07,680 traditional industries. How how does how do you see that 837 00:53:07,740 --> 00:53:08,940 impacting the economy? 838 00:53:09,209 --> 00:53:12,629 Yeah, I think that it's um you probably feel it in in the 839 00:53:12,989 --> 00:53:17,069 day to day work that you do too. Like the the costs of producing 840 00:53:17,159 --> 00:53:21,079 podcasts are are going down because um you know the cost of 841 00:53:21,499 --> 00:53:23,929 getting everything, like a microphone, or and all the 842 00:53:23,929 --> 00:53:29,059 softwares. There are a lot better technological advances as 843 00:53:29,119 --> 00:53:33,049 relates to software engineering and hardware manufacturing. I 844 00:53:33,139 --> 00:53:38,569 think that yeah, like looking at like a restaurant, there's going 845 00:53:38,569 --> 00:53:44,259 to be a lot better ways to for the food service distributors to 846 00:53:44,469 --> 00:53:48,249 actually understand where the inks are in their supply chains, 847 00:53:48,369 --> 00:53:51,909 where there is inventory over time, and really drive down that 848 00:53:51,969 --> 00:53:56,979 cost over time. You know, AI will allow for autonomous truck 849 00:53:57,039 --> 00:54:01,409 drivers that so that will decline the amount of people who 850 00:54:01,529 --> 00:54:04,709 actually have to drive the truck from a place A to B and 851 00:54:05,489 --> 00:54:08,519 necessarily drive down the costs there. I think that there are 852 00:54:08,609 --> 00:54:12,179 just so many different ways that it's going to accelerate things 853 00:54:12,299 --> 00:54:17,849 and it's going to touch every part of our economy, especially 854 00:54:17,909 --> 00:54:21,409 every part that uses any sort of intelligence on a day-to-day 855 00:54:21,529 --> 00:54:24,709 basis. If you're not really just going through the motions, then 856 00:54:24,979 --> 00:54:28,009 it's going to drive down costs significantly. And even if you 857 00:54:28,129 --> 00:54:30,679 are just going through the motions, it will allow you to 858 00:54:31,909 --> 00:54:37,009 really think about new ways and to do the job more efficiently, 859 00:54:37,189 --> 00:54:41,229 implement new technologies, etc. I'm pretty bullish on just the 860 00:54:41,559 --> 00:54:44,979 productivity of any country out there with these tools, 861 00:54:46,000 --> 00:54:48,580 yeah, and I think this is a great place to 862 00:54:48,580 --> 00:54:52,810 try and to wind this down. But there's just so much to unpack 863 00:54:52,900 --> 00:54:56,410 in here. But Liam, if if someone listening is going to take away 864 00:54:56,620 --> 00:54:59,800 one idea from the conversation or from your research, what 865 00:54:59,830 --> 00:55:01,020 would. What would you like that to be? 866 00:55:01,620 --> 00:55:07,350 Yeah, I would say really try to understand the how this 867 00:55:07,740 --> 00:55:10,530 is a lot more of an exponential technology than anything that 868 00:55:10,530 --> 00:55:16,020 we've ever seen, including the internet. The just look back at 869 00:55:16,050 --> 00:55:19,080 that chart of how many tokens Open Rotary is doing of a 870 00:55:19,740 --> 00:55:24,560 25,000% increase-that's just astronomical-and there are going 871 00:55:24,560 --> 00:55:27,560 to be so many opportunities that are created out of this. Whether 872 00:55:27,620 --> 00:55:31,820 it's upleveling your own skills, creating new businesses out of 873 00:55:31,850 --> 00:55:36,290 this, I think it's just-I would highly recommend reading this 874 00:55:36,350 --> 00:55:41,260 report to see where, at least I think, value will accrue in this 875 00:55:41,890 --> 00:55:46,480 AI native world, and where it may give you ideas for starting 876 00:55:46,570 --> 00:55:49,300 businesses, et cetera, and really encourage founders who 877 00:55:49,390 --> 00:55:53,800 are on the cutting edge to reach out to us here. And in addition, 878 00:55:53,890 --> 00:55:56,920 too, I would just think about how to really become the most AI 879 00:55:57,220 --> 00:56:00,720 version of yourself because it will only bring you a lot of 880 00:56:00,720 --> 00:56:02,040 benefits in the future. 881 00:56:02,160 --> 00:56:04,920 I know research is a really core part of what 882 00:56:05,010 --> 00:56:08,250 Early Writers does or provides to the ecosystem. You guys have 883 00:56:08,340 --> 00:56:11,430 some great newsletters that I'd encourage everyone to sign up 884 00:56:11,520 --> 00:56:15,210 for. But Singularity Stack is not just, as I understand it, a 885 00:56:15,300 --> 00:56:18,150 one-off report, but you you you have a living part of your 886 00:56:18,180 --> 00:56:21,320 website that will continually update. Can you talk a little 887 00:56:21,350 --> 00:56:24,470 bit about how people can access what you've already done, and 888 00:56:24,470 --> 00:56:27,410 then how to stay on top of the developments that you see since 889 00:56:27,530 --> 00:56:29,180 you've published it? 890 00:56:29,329 --> 00:56:32,689 Yeah, for sure. We, yeah, please check out 891 00:56:32,869 --> 00:56:36,019 earlywriters.com. We publish a bunch of research there. You can 892 00:56:36,109 --> 00:56:39,079 subscribe to our newsletter there too, and then we also have 893 00:56:39,679 --> 00:56:43,569 a weekly newsletter that just goes into that tries to track 894 00:56:43,629 --> 00:56:46,419 the singularity over time, and the singularity, if if we 895 00:56:46,479 --> 00:56:48,999 haven't already mentioned it, it's just defined as the 896 00:56:49,809 --> 00:56:53,829 acceleration that is too great for the human eye to really 897 00:56:53,919 --> 00:56:57,519 measure. And so, while while we can't necessarily measure each 898 00:56:57,579 --> 00:57:00,519 single improvement over time, we're trying to really keep 899 00:57:00,609 --> 00:57:03,779 everybody abreast to the latest developments as it relates to 900 00:57:03,869 --> 00:57:06,899 open source advances, what it means for the economy and 901 00:57:07,169 --> 00:57:11,609 Bitcoin and companies in the space. And so, if you subscribe 902 00:57:11,699 --> 00:57:16,319 to our website, we'll add you to all of our research lists. And 903 00:57:17,429 --> 00:57:19,859 please reach out if you're building companies on the edge 904 00:57:19,949 --> 00:57:23,959 of this too. liam@earlygriders.com, and 905 00:57:24,109 --> 00:57:25,819 really appreciate you having me on today, Lynne. 906 00:57:26,929 --> 00:57:28,969 We thank you for your research. We thank you for 907 00:57:28,999 --> 00:57:31,699 all the thought that you're giving this, and the ideas for 908 00:57:31,759 --> 00:57:36,139 opportunities, both for founders as well as for investors. And it 909 00:57:36,199 --> 00:57:39,139 just gives a lot to think about in a positive way. That I think 910 00:57:39,619 --> 00:57:43,089 you know, with a narrative that is becoming more fearful toward 911 00:57:43,539 --> 00:57:46,329 AI, it's so important to look at what benefits it's achieving 912 00:57:46,509 --> 00:57:50,889 already and what the potential is for the just the advancement 913 00:57:51,249 --> 00:57:54,219 of of where we go from here. So we we really appreciate it. 914 00:57:54,279 --> 00:57:57,219 Thank you for listening to this episode of Build with Bitcoin. 915 00:57:57,369 --> 00:58:00,479 If you found it of value, please take a second to like, 916 00:58:00,569 --> 00:58:03,659 subscribe, or share, this helps the visibility and reach of our 917 00:58:03,749 --> 00:58:06,269 podcast, so that more people are aware of the innovation 918 00:58:06,359 --> 00:58:10,259 happening in and around Bitcoin. It truly means a lot. I'd also 919 00:58:10,319 --> 00:58:12,989 like to take this opportunity to invite you to join us at an 920 00:58:13,049 --> 00:58:16,949 event that Israel and I are hosting, Meeta Tech Talks 2026, 921 00:58:17,519 --> 00:58:21,469 taking place from October 25th to the 27th at the exclusive 922 00:58:21,649 --> 00:58:25,099 Puntamita Resort in Mexico. This will be an intimate gathering of 923 00:58:25,609 --> 00:58:29,689 125 invite-only participants, exploring how the future of 924 00:58:29,749 --> 00:58:33,439 capital is being shaped by the intersection of Bitcoin, 925 00:58:33,799 --> 00:58:37,459 artificial intelligence, and energy. To request an invite or 926 00:58:37,519 --> 00:58:41,829 for more information, visit mitatechtalks.com. We hope 927 00:58:41,889 --> 00:58:45,099 you'll join us, and thanks again for being a part of the Build 928 00:58:45,129 --> 00:58:46,179 with Bitcoin community.