WEBVTT 00:00:00.160 --> 00:00:03.677 this new wave of AI companies is is growing revenue like just like actual 00:00:03.760 --> 00:00:07.117 customer revenue, actual demand translated through to dollars showing up 00:00:07.200 --> 00:00:10.237 in bank accounts at like an absolutely unprecedented takeoff rate. We're seeing 00:00:10.320 --> 00:00:14.237 companies grow much faster. I'm very skeptical that the form and shape of the 00:00:14.320 --> 00:00:16.557 products that people are using today is what they're going to be using in 5 or 00:00:16.640 --> 00:00:19.037 10 years. I think things are going to get much more sophisticated from here. 00:00:19.120 --> 00:00:21.757 And so I think we probably have a long way to go. These are trillion dollar 00:00:21.840 --> 00:00:25.117 questions, not answers. But once somebody proves that it's capable, it 00:00:25.200 --> 00:00:28.317 seems to not be that hard for other people to be able to catch up, even 00:00:28.400 --> 00:00:32.397 people with far less resources. When a company is confronted with fundamentally 00:00:32.480 --> 00:00:36.797 open strategic or economic questions, it's often a big problem. Companies like 00:00:36.880 --> 00:00:39.197 need to answer these questions and if they get the answers wrong, they're 00:00:39.280 --> 00:00:42.797 really in trouble. Venture, we can bet on multiple strategies at the same time. 00:00:42.880 --> 00:00:46.077 We are aggressively investing behind every strategy that we've identified 00:00:46.160 --> 00:00:48.957 that we think has a plausible chance of working. If you want to understand 00:00:49.040 --> 00:00:51.677 people, there's basically two ways to understand what people are doing and 00:00:51.760 --> 00:00:55.597 thinking. One is to ask them and then the other is to watch them. And what you 00:00:55.680 --> 00:00:58.957 often see in many areas of human activity, including politics and many 00:00:59.040 --> 00:01:01.917 different aspects of society, the answers that you get when you ask people 00:01:02.000 --> 00:01:04.557 are very different than the answers that you get when you watch them. If you run 00:01:04.640 --> 00:01:08.877 a survey or a poll of what, for example, American voters think about AI, it's 00:01:08.960 --> 00:01:10.077 just like they're all in a total panic. 00:01:10.160 --> 00:01:11.437 It's like, "Oh my god, this is terrible. 00:01:11.520 --> 00:01:13.197 This is awful. It's going to kill all the jobs. It's going to ruin 00:01:13.280 --> 00:01:19.040 everything." If you watch the revealed preferences, they're all using AI. 00:01:22.320 --> 00:01:25.197 A lot of folks have sent questions ahead of time and and what I what I've done is 00:01:25.280 --> 00:01:28.957 kind of curated into a few different sections uh in in an AMA this morning 00:01:29.040 --> 00:01:32.877 with uh with Mark. So, what we thought we'd do is cover uh four big topics. So, 00:01:32.960 --> 00:01:36.957 AI and what's happening in the markets, policy and regulation, um all things 00:01:37.040 --> 00:01:40.717 816Z, and then we've got a a fun catchall which we're we're calling 00:01:40.800 --> 00:01:44.637 sandbox of things if we get to it. So, starting first maybe with uh with the 00:01:44.720 --> 00:01:48.077 biggest question. We're sitting in the middle of the AI revolution. Mark, what 00:01:48.160 --> 00:01:51.197 inning do you think we're in and and what are you most excited about? 00:01:51.280 --> 00:01:54.397 >> First of all, I I would say this is the biggest tech technological revolution of 00:01:54.480 --> 00:01:58.797 my life. Um and you know, hopefully I'll see more like this in the next whatever 00:01:58.880 --> 00:02:02.637 30 years, but I I mean this is the big one. Um and just in terms of order of 00:02:02.720 --> 00:02:06.557 magnitude, like this is clearly bigger than the internet. Um like the the the 00:02:06.640 --> 00:02:09.917 comps on this are things like the microprocessor and the steam engine and 00:02:10.000 --> 00:02:11.757 electricity. So that this is a really 00:02:11.840 --> 00:02:14.797 this is a really big one. um the wheel. 00:02:14.880 --> 00:02:19.197 Um the reason this is so big, I mean maybe obvious to folks at this point, 00:02:19.280 --> 00:02:23.197 but I'll just go through it quickly. So um if you kind of trace all the way back 00:02:23.280 --> 00:02:26.237 to the 1930s, uh there's a great book called Rise of the Machines that kind of 00:02:26.320 --> 00:02:29.037 goes through this. Um if you trace all the way back to the 1930s, there was 00:02:29.120 --> 00:02:32.477 actually a debate among the people who actually invented the computer. Um and 00:02:32.560 --> 00:02:35.757 it was this this sort of debate between whether computer they kind of understood 00:02:35.840 --> 00:02:38.637 the theory of computation before they before they they actually built the 00:02:38.720 --> 00:02:42.557 things. Um and um they they had this big debate over whether the computer should 00:02:42.640 --> 00:02:45.757 be basically built in the image of what at the time were called adding machines 00:02:45.840 --> 00:02:49.117 or calculating machines where you know think of sort of essentially cash 00:02:49.200 --> 00:02:52.877 registers. Um IBM is actually the successor company to the national cash 00:02:52.960 --> 00:02:57.117 register company uh of America. Um and so like and and and that was of course 00:02:57.200 --> 00:03:00.317 the path that the industry took which was building these kind of hyper literal 00:03:00.400 --> 00:03:03.757 you know mathematical machines you know that could execute mathematical 00:03:03.840 --> 00:03:06.877 operations billions of times per second but of course had no ability to kind of 00:03:06.960 --> 00:03:09.997 deal with human beings the way humans like to be dealt with and so you know 00:03:10.080 --> 00:03:13.597 couldn't understand you know human speech human language um and so forth 00:03:13.680 --> 00:03:16.317 and and that's the computer industry that got built over the last 80 years 00:03:16.400 --> 00:03:19.437 and that's the computer industry that's built all the wealth of uh uh and and 00:03:19.520 --> 00:03:22.797 financial returns of the computer industry uh over the last 80 years you 00:03:22.880 --> 00:03:25.757 know across all the generations of computers from mainframes through to 00:03:25.840 --> 00:03:29.357 smartphones. Um but but they knew at the time they knew in the 30s actually they 00:03:29.440 --> 00:03:32.077 understood the basic structure of the human brain. They understood they had a 00:03:32.160 --> 00:03:35.117 theory of sort of human cognition and and and actually they had the theory of 00:03:35.200 --> 00:03:39.677 neural networks. Um and so they they had this theory that um the there's actually 00:03:39.760 --> 00:03:43.837 the first neural network uh paper academic paper was published in 1943 you 00:03:43.920 --> 00:03:48.957 know which was over 80 years ago which is extremely amazing. Um there's an 00:03:49.040 --> 00:03:51.357 interview you can read an interview or you can watch an interview on YouTube 00:03:51.440 --> 00:03:54.957 with uh these two authors Makulla and Pitts and you can watch an interview I 00:03:55.040 --> 00:03:59.197 think with Makulla on YouTube from like I don't know 1946 or something. He was 00:03:59.280 --> 00:04:02.477 like on TV you know in the in the ancient past and it's literally like 00:04:02.560 --> 00:04:04.797 it's amazing interview because it's like him in his beach house and for some 00:04:04.880 --> 00:04:08.317 reason he's not wearing a shirt um and he's like you know talking about like 00:04:08.400 --> 00:04:11.117 this future in which computers are going to be you know built on on the model of 00:04:11.200 --> 00:04:15.037 a human brain through through neural networks. Um and and that was the path 00:04:15.120 --> 00:04:18.077 not taken. And basically what happened was right the computer industry got 00:04:18.160 --> 00:04:21.917 built in the in the image of of like the adding machine. Um but and the neural 00:04:22.000 --> 00:04:25.117 network basically didn't happen but the neural network as an idea continued to 00:04:25.200 --> 00:04:29.357 be explored in academia um and sort of advanced research by sort of a rump you 00:04:29.440 --> 00:04:32.717 know movement that was originally called cybernetics and then became known as as 00:04:32.800 --> 00:04:36.157 artificial intelligence uh basically for the last 80 years and and and 00:04:36.240 --> 00:04:39.037 essentially it didn't work like essentially it was basically decade 00:04:39.120 --> 00:04:43.117 after decade after decade of excessive optimism uh followed by disappointment. 00:04:43.200 --> 00:04:47.117 When I was in college in the 80s, there had been a famous kind of AI boom bust 00:04:47.200 --> 00:04:51.437 uh cycle in the 80s in venture and in Silicon Valley. Um I mean it was tiny by 00:04:51.520 --> 00:04:55.437 by modern standards, but it it at the time was a big deal. Um and um you know 00:04:55.520 --> 00:04:59.357 by the time I got to college in '89 um in computer science departments, AI was 00:04:59.440 --> 00:05:01.757 kind of a backwater field and everybody kind of assumed that it was never going 00:05:01.840 --> 00:05:04.637 to happen. But the scientists kept working on it to their credit and they 00:05:04.720 --> 00:05:08.237 they they built up this kind of enormous reservoir of of concepts and ideas and 00:05:08.320 --> 00:05:12.557 then basically we all saw what happened with the CHIGPT uh moment. all of a 00:05:12.640 --> 00:05:15.677 sudden it it sort of crystallized. It's like oh my god, right? It turns out it 00:05:15.760 --> 00:05:19.117 works. Um and and so you know that that's the moment we're in now. And then 00:05:19.200 --> 00:05:21.757 you know really significantly that was what you that was less than three years 00:05:21.840 --> 00:05:26.157 ago, right? That was the summer of 20 it was the the Christmas of 22. So, we're 00:05:26.240 --> 00:05:30.637 sort of three year we're we're sort of three years in um to, you know, 00:05:30.720 --> 00:05:34.957 basically what is effectively effectively an 80-year revolution um of 00:05:35.040 --> 00:05:38.077 actually being able to deliver on all the promise that the that the people on 00:05:38.160 --> 00:05:41.517 the all the on the alternate path, the sort of human cognition model path, you 00:05:41.600 --> 00:05:44.157 know, kind of saw from the very beginning and and then, you know, the 00:05:44.240 --> 00:05:47.757 great news with this technology is it's already it's kind of ultra democratized. 00:05:47.840 --> 00:05:50.797 you know, the best AI in the world is available. Launch at GPD or Grock or 00:05:50.880 --> 00:05:53.837 Gemini or or um you know, these other you know, these other products that you 00:05:53.920 --> 00:05:57.597 can just use um and you can just kind of see how they work and you know, same 00:05:57.680 --> 00:06:01.357 thing for video, you can see with Sora and VO kind of state-of-the-art uh with 00:06:01.440 --> 00:06:05.277 that you can see with music, you can see you know uh Suno and IDO and so forth. 00:06:05.360 --> 00:06:09.437 Um and so like you know we're basically seeing that happen and now and now 00:06:09.520 --> 00:06:13.997 Silicon Valley is responding with this just like incredible rush of enthusiasm. 00:06:14.080 --> 00:06:16.957 And you know, really critically this gets to the magic of Silicon Valley, 00:06:17.040 --> 00:06:20.477 which is, you know, Silicon Valley long since has ceased to be a place where 00:06:20.560 --> 00:06:23.517 people make silicon that, you know, that's that not long ago moved out out 00:06:23.600 --> 00:06:26.637 of the out out of California and then ultimately out of the US, although we're 00:06:26.720 --> 00:06:30.477 trying to bring it back now. Um but but but the great kind of virtue of Silicon 00:06:30.560 --> 00:06:34.077 Valley o over the last you know over the last you know 80 years of its existence 00:06:34.160 --> 00:06:38.077 is its ability to kind of uh recycle talent from previous waves of technology 00:06:38.160 --> 00:06:41.517 and new waves of technology uh and then inspire an entire new generation of 00:06:41.600 --> 00:06:45.277 talent you know to basically come join the you know join the project. Um and so 00:06:45.360 --> 00:06:47.997 Silicon Valley has this recurring pattern of being able to reallocate 00:06:48.080 --> 00:06:50.797 capital and talent and build enthusiasm and build critical mass and build 00:06:50.880 --> 00:06:53.997 funding support and build you know human capital and build you know everything 00:06:54.080 --> 00:06:56.877 enthusiasm um you know for each new wave 00:06:56.960 --> 00:07:01.517 of technology. So, so that's what's happening with AI. Um, you know, I I 00:07:01.600 --> 00:07:04.477 think probably the biggest thing I could just say is like I'm surprised I think 00:07:04.560 --> 00:07:08.637 essentially on a daily basis of what I'm seeing. Um, uh, and and you know, we 00:07:08.720 --> 00:07:11.597 we're we're in the fortunate position to kind of get to see it from from two 00:07:11.680 --> 00:07:15.517 angles. Uh, you know, one one is we track the underlying science and and, 00:07:15.600 --> 00:07:18.157 uh, and kind of research work very carefully. And so I would say like every 00:07:18.240 --> 00:07:22.317 day I see a new AI research paper that just like completely floores me um of 00:07:22.400 --> 00:07:25.677 some new capability um or some new discovery uh or some new development 00:07:25.760 --> 00:07:28.877 that I that I would have never anticipated that I I'm just like wow I 00:07:28.960 --> 00:07:31.757 you know I can't believe this is happening. And then um on the other side 00:07:31.840 --> 00:07:34.957 of course you know we see the flow of all of the new uh products uh and all 00:07:35.040 --> 00:07:39.597 the new startups. Um and you know I would say we're routinely um you know 00:07:39.680 --> 00:07:42.957 kind of seeing things and again kind of have my my jaw on the floor. Um, and so, 00:07:43.040 --> 00:07:47.677 you know, it feels like we we we've unlocked this giant vista. Um, I do 00:07:47.760 --> 00:07:50.877 think it's going to kind of come in fits and starts. Um, you know, the these 00:07:50.960 --> 00:07:54.797 things are messy processes. Um, you know, you know, this is an industry that 00:07:54.880 --> 00:07:58.477 kind of routinely gets out over risks and overpromises. Um, and and so, you 00:07:58.560 --> 00:08:00.717 know, there, you know, there will certainly be points where it's like, 00:08:00.800 --> 00:08:03.117 wow, you know, this isn't working as well as people thought, or you know, 00:08:03.200 --> 00:08:05.597 wow, this turns out to be too expensive and the economics don't work or 00:08:05.680 --> 00:08:08.397 whatever. But, you know, against that, I would just say the capabilities are 00:08:08.480 --> 00:08:11.277 truly magical. And and by the way, I think that's the experience that 00:08:11.360 --> 00:08:12.557 consumers are having when they use it. 00:08:12.640 --> 00:08:15.037 And I think that's the experience that businesses are having for the most part 00:08:15.120 --> 00:08:18.237 when they uh you know, when when they're working on their pilots and and looking 00:08:18.320 --> 00:08:21.757 at adoption and and and then and then it translates to the underlying numbers. I 00:08:21.840 --> 00:08:24.957 mean, we're we're just seeing a this new wave of AI companies is is growing 00:08:25.040 --> 00:08:29.117 revenue like just like actual customer revenue, actual demand uh translated 00:08:29.200 --> 00:08:32.637 through to dollars showing up in bank accounts. Um you know, at like an 00:08:32.720 --> 00:08:34.077 absolutely unprecedented takeoff rate. 00:08:34.160 --> 00:08:36.237 We're seeing companies grow much faster. 00:08:36.320 --> 00:08:39.597 um uh you the the key leading AI companies and the companies that have 00:08:39.680 --> 00:08:42.637 real breakthroughs um and have real have very compelling products are growing 00:08:42.720 --> 00:08:45.597 revenues that you know kind of faster than any any way I've certainly ever 00:08:45.680 --> 00:08:49.277 seen before. Um and so like just just from all that it kind of feels like it 00:08:49.360 --> 00:08:53.357 has to be early. Like it it's kind of hard to imagine that we've like we we've 00:08:53.440 --> 00:08:56.317 topped out in any way. It feels like everything is still developing. I mean 00:08:56.400 --> 00:08:59.117 quite frankly it feels like the products to me it feels like the products are 00:08:59.200 --> 00:09:03.437 still super early. Like I'm I'm I'm very skeptical that the form and shape of the 00:09:03.520 --> 00:09:05.517 products that people are using today is what they're going to be using in five 00:09:05.600 --> 00:09:08.157 or 10 years. I think I think things are going to get much more sophisticated 00:09:08.240 --> 00:09:11.037 from here. Um and so I think we probably have a long way to go. 00:09:11.120 --> 00:09:15.357 >> Maybe on that that topic. So one of the big knocks is yes the revenue is immense 00:09:15.440 --> 00:09:19.597 but the expenses seem to also be keeping pace. So like what are people missing as 00:09:19.680 --> 00:09:21.597 a part of that discussion and topic? 00:09:21.680 --> 00:09:25.517 >> Yeah. So just start with just like core business models, right? Um and so you're 00:09:25.600 --> 00:09:28.157 right. There's basically this industry basically has two two core business 00:09:28.240 --> 00:09:31.437 models. consumer business model and the quote unquote enterprise uh or 00:09:31.520 --> 00:09:35.037 infrastructure business model. Um you know look on the on the consumer side we 00:09:35.120 --> 00:09:38.317 we just live in a very interesting world now where where the internet exists and 00:09:38.400 --> 00:09:42.157 is fully deployed right. Um, and so I'll give you an example. Sometimes people 00:09:42.240 --> 00:09:44.717 ask us like, "Is AI like the internet revolution?" It's like, well, a little 00:09:44.800 --> 00:09:47.197 bit, but like the thing with the internet was we had to build the 00:09:47.280 --> 00:09:51.357 internet. Like we we like we had we had to actually build the network and we 00:09:51.440 --> 00:09:53.917 actually had to, you know, and ultimately it involved enormous amount 00:09:54.000 --> 00:09:56.637 of fiber in the ground and it involved enormous numbers of like mobile cell 00:09:56.720 --> 00:09:59.757 towers and, you know, enormous number of, you know, shipments of of of 00:09:59.840 --> 00:10:03.197 smartphones and tablets and and and laptops in order to get people on the 00:10:03.280 --> 00:10:06.637 internet. Like there was this like just like incredible physical lift um, you 00:10:06.720 --> 00:10:09.837 know, to do that. And and by the way, people forget how long that took. Uh 00:10:09.920 --> 00:10:12.957 right, the the the you know, the internet itself is a invention of the 00:10:13.040 --> 00:10:17.517 1960s, 1970s. Um the consumer internet, you know, was a new phenomenon in the 00:10:17.600 --> 00:10:21.277 early '90s. Um but, you know, we didn't really get broadband to the home until 00:10:21.360 --> 00:10:24.397 the 2000s. You know, that really didn't start rolling out actually until after 00:10:24.480 --> 00:10:26.317 the com crash, which is fairly amazing. 00:10:26.400 --> 00:10:30.397 And then we didn't get mobile broadband until like 2010. And and people actually 00:10:30.480 --> 00:10:32.397 forget the original iPhone dropped in 00:10:32.480 --> 00:10:34.877 2007. It didn't have broadband. it was 00:10:34.960 --> 00:10:37.757 on a it was on a narrowband 2G network. 00:10:37.840 --> 00:10:40.877 Um it did not have high speed like it did not have anything resembling 00:10:40.960 --> 00:10:44.477 high-speed data. Um and so it wasn't really until you know really about 15 00:10:44.560 --> 00:10:47.757 years ago that we even had mobile broadband. So so the internet was this 00:10:47.840 --> 00:10:50.957 massive lift but but the internet got built right and smartphones 00:10:51.040 --> 00:10:54.157 proliferated. And so the point is now you have 5 billion people on planet 00:10:54.240 --> 00:10:58.157 earth that are on some version of you know mobile broadband internet right um 00:10:58.240 --> 00:11:00.717 and you know smartphones all over the world are selling for you know as little 00:11:00.800 --> 00:11:04.317 as like 10 bucks. Um and you know you have these you know amazing projects 00:11:04.400 --> 00:11:07.437 like geo and India that are bringing you know you know the sort of the remaining 00:11:07.520 --> 00:11:09.917 you know kind of the remaining population of of planet earth that 00:11:10.000 --> 00:11:13.277 hasn't been online until now is coming online. So, you know, so we're talking 00:11:13.360 --> 00:11:17.037 five billion, six billion, you know, people and and then the consumer, the 00:11:17.120 --> 00:11:19.917 reason I go through that is the consumer AI products could basically deploy to 00:11:20.000 --> 00:11:24.237 all of those people basically as quickly as they want to adopt, right? Um, and so 00:11:24.320 --> 00:11:27.997 sort of the internet's the carrier wave for AI to be able to proliferate at kind 00:11:28.080 --> 00:11:31.917 of light speed uh uh into the broad base of the global population. And and that's 00:11:32.000 --> 00:11:35.277 a let's just say that's a potential rate of proliferation of a new technology 00:11:35.360 --> 00:11:38.317 that's just far faster than has ever been possible before. Like what you 00:11:38.400 --> 00:11:41.677 know, like you couldn't download electricity, right? you you couldn't 00:11:41.760 --> 00:11:45.517 download, you know, you couldn't download indoor plumbing. Um, you know, 00:11:45.600 --> 00:11:48.877 you couldn't download television, but you can download AI. Um, and and and 00:11:48.960 --> 00:11:52.077 this is what we're seeing, which is the AI consumer, you know, the AI consumer 00:11:52.160 --> 00:11:56.317 killer applications are growing at at at an incredible rate. Um, and then and 00:11:56.400 --> 00:11:59.677 then they're monetizing really well. Um, and and again, you know, we we I 00:11:59.760 --> 00:12:02.077 mentioned this already, but like generally speaking, the monetization is 00:12:02.160 --> 00:12:06.157 is very good. Um, by the way, including at higher price points. Um, one of the 00:12:06.240 --> 00:12:09.437 things I like about the um, you know, about watching the AI wave is the AI 00:12:09.520 --> 00:12:12.477 companies I think are are more creative on pricing than the SAS companies and 00:12:12.560 --> 00:12:13.517 the consumer internet companies were. 00:12:13.600 --> 00:12:16.957 And so it's it's for example now becoming routine to have $200 or $300 t 00:12:17.040 --> 00:12:20.237 per month tiers uh, for consumer AI which I which I think is very positive 00:12:20.320 --> 00:12:25.037 because I I think the I think a lot of companies cap their kind of opportunity 00:12:25.120 --> 00:12:28.157 by by capping their pricing uh, kind of too low and I think the AI companies are 00:12:28.240 --> 00:12:31.997 more willing to push that which I think is good. So anyway, so that you know I 00:12:32.080 --> 00:12:35.117 think that's reason for like I would say you know considerable rational optimism 00:12:35.200 --> 00:12:37.917 for the scope of of consumer revenues that we're going to be talking about 00:12:38.000 --> 00:12:41.117 here. And then on the enterprise side, you know, there the question is 00:12:41.200 --> 00:12:45.517 basically just, you know, what is intelligence worth, right? Um, and you 00:12:45.600 --> 00:12:48.157 know, if you have the ability to like inject more intelligence into your 00:12:48.240 --> 00:12:51.437 business and you have the ability to do, you know, even the most prosaic things 00:12:51.520 --> 00:12:55.437 like raise your customer service scores, uh, you know, increase upsells, um, uh, 00:12:55.520 --> 00:12:58.317 you know, or reduce churn or if you have the ability to, um, you know, run 00:12:58.400 --> 00:13:01.437 marketing campaigns more effectively, um, you know, all of which AI is 00:13:01.520 --> 00:13:04.717 directly relevant to, like, you know, these are like direct business payoffs, 00:13:04.800 --> 00:13:07.677 um, you know, that people are seeing already. Um, and then if you have the 00:13:07.760 --> 00:13:11.117 opportunity to infuse AI into new products and all of a sudden, you know, 00:13:11.200 --> 00:13:14.157 all of a sudden your car talks to you, um, and everything in the world kind of 00:13:14.240 --> 00:13:17.197 lights up and starts to get really smart. Um, you know, you know, what's 00:13:17.280 --> 00:13:20.077 that worth? And and again there you just you you kind of observe it and you're 00:13:20.160 --> 00:13:22.717 like, wow, the the leading AI infrastructure companies are growing 00:13:22.800 --> 00:13:26.797 revenues incredibly quickly. Um, you know, the pull is really tremendous. Um, 00:13:26.880 --> 00:13:30.877 and so, you know, again there it's just it feels like this just like incredible 00:13:30.960 --> 00:13:34.477 uh, you know, product market fit. Um and and and the core business model, right, 00:13:34.560 --> 00:13:37.437 is is is actually quite quite interesting. The core business model is 00:13:37.520 --> 00:13:41.117 is is basically is basically tokens by the drink, right? And so it's it's sort 00:13:41.200 --> 00:13:44.957 of tokens of intelligence uh you know, per dollar. And oh, and then by the way, 00:13:45.040 --> 00:13:47.837 this is the other fun thing is if you look at what's happening with uh the 00:13:47.920 --> 00:13:53.437 price of AI, the price of AI is falling much faster than Moore's law. And when I 00:13:53.520 --> 00:13:56.157 could go through that in great detail, but basically like all of the inputs 00:13:56.240 --> 00:14:01.117 into AI on a perunit basis, the costs are collapsing. Um and and and and then 00:14:01.200 --> 00:14:04.237 as a consequence there's kind of this hyperdelation of per unit cost and then 00:14:04.320 --> 00:14:08.237 that is like driving you know just like you know a more than corresponding level 00:14:08.320 --> 00:14:12.637 of demand growth you know with with with elasticity. Um and so you know even 00:14:12.720 --> 00:14:15.597 there we're like it feels like we're just at the very beginning of kind of 00:14:15.680 --> 00:14:18.797 you know figuring out exactly how you know expensive or cheap this stuff is 00:14:18.880 --> 00:14:21.357 getting. I mean look there's just no question tokens by the drink are going 00:14:21.440 --> 00:14:24.077 to get a lot cheaper from here. Um that's just going to drive I think 00:14:24.160 --> 00:14:27.437 enormous demand. Um and then everything in the cost structure is going to get 00:14:27.520 --> 00:14:31.757 optimized right? Um and so you know when when people talk about like you know the 00:14:31.840 --> 00:14:35.757 chips or you know whatever you know kind of the unit input costs for building AI 00:14:35.840 --> 00:14:40.077 you know you now have these like m the losses of blind demand are are going to 00:14:40.160 --> 00:14:44.077 are going to kick in right um what's the you know in any market that has sort of 00:14:44.160 --> 00:14:47.357 commodity like characteristics you know the number one cause of a of a of of a 00:14:47.440 --> 00:14:50.717 glut is a shortage and the number one cause of a shortage is the glut right um 00:14:50.800 --> 00:14:54.637 and so you have you know to the extent you have like shortage of GPUs or 00:14:54.720 --> 00:14:57.517 shortage of whatever infest chips or shortage of you know whatever data 00:14:57.600 --> 00:15:01.277 center case, you know, if you look at just the history of humanity building 00:15:01.360 --> 00:15:04.557 things in response to demand, you know, if there's a shortage of something that 00:15:04.640 --> 00:15:08.877 can be physically replicated, it it does get replicated. Um, and so there's going 00:15:08.960 --> 00:15:11.357 to be like just enormous build out of all I mean there is there's just 00:15:11.440 --> 00:15:14.077 hundreds of billions or at this point trillions of dollars maybe going into 00:15:14.160 --> 00:15:17.837 the ground um in all these things. And so the the per unit cost of the AI 00:15:17.920 --> 00:15:19.677 companies are going to drop like a rock 00:15:19.760 --> 00:15:24.877 um you know over the course of the next decade. Um and so like I yeah I mean the 00:15:24.960 --> 00:15:27.357 economic questions of course are very real and of course there's you know 00:15:27.440 --> 00:15:30.557 microeconomic questions around around all these businesses but the the sort of 00:15:30.640 --> 00:15:34.477 macro forces have been at least here I think are very strong um and and yeah I 00:15:34.560 --> 00:15:37.197 I just given the underlying value of the 00:15:37.280 --> 00:15:41.597 of of this technology to both the consumers the enterprise users. Um, and 00:15:41.680 --> 00:15:44.717 given the just like incredibly aggressive discovery that's happening of 00:15:44.800 --> 00:15:47.437 of all the ways that people can use this in their lives and in their businesses, 00:15:47.520 --> 00:15:50.157 like it's just it's really hard for me to see how it both doesn't grow a lot 00:15:50.240 --> 00:15:51.997 and generate just enormous revenue. 00:15:52.080 --> 00:15:55.277 >> Yeah. And actually, I think it was like two or three weeks ago where AWS was 00:15:55.360 --> 00:15:58.077 saying like the the GPUs that they've been using, they've been able to extend 00:15:58.160 --> 00:15:59.917 back to even like seven plus years. So 00:16:00.000 --> 00:16:02.317 like the shelf life also of the GPUs 00:16:02.400 --> 00:16:05.117 that they're using is now extending in 00:16:05.200 --> 00:16:09.757 ways of which they can optimize better than maybe perhaps the last couple of of 00:16:09.840 --> 00:16:12.637 cycles. as well. Is that the right way to think about it as well? 00:16:12.720 --> 00:16:15.277 >> Yeah, that's right. And then and then that's one that's that's one really 00:16:15.360 --> 00:16:18.317 important question and observation and and then by the way that also gets to 00:16:18.400 --> 00:16:21.837 this other kind of question um where there's different theories on it. Um 00:16:21.920 --> 00:16:24.317 which is basically big models versus small models. 00:16:24.400 --> 00:16:28.157 >> Um and so a a lot of the data a lot of the data center build is oriented around 00:16:28.240 --> 00:16:31.677 hosting um training and and and and serving the the big the big models, you 00:16:31.760 --> 00:16:33.357 know, for for all the obvious reasons. 00:16:33.440 --> 00:16:36.637 Um but there's also the small the small model revolution is happening at the 00:16:36.720 --> 00:16:39.597 same time and and and and if you just kind of track you know you can get get 00:16:39.680 --> 00:16:42.877 the various research firms have these charts you can get um but if you just 00:16:42.960 --> 00:16:45.437 kind of track the if you track the capability of the leading edge models 00:16:45.520 --> 00:16:48.397 over time what you find is after 6 or 12 months there's a small model that's just 00:16:48.480 --> 00:16:52.477 as capable um and so there there there's this kind of chase function that's 00:16:52.560 --> 00:16:55.517 happening which is the capabilities of the big models are basically being 00:16:55.600 --> 00:17:00.157 shrunk shrunk down uh and provided at at at at smaller size and then therefore 00:17:00.240 --> 00:17:03.357 smaller cost you know quite quickly. So, I I'll just give you the the most recent 00:17:03.440 --> 00:17:05.677 example that just got hit over the last two weeks. And again, this is a thing 00:17:05.760 --> 00:17:09.757 that's just kind of shocking. Um is there's this Chinese company that has a 00:17:09.840 --> 00:17:12.237 um well, I forget the name of the company, but it's it's uh the company 00:17:12.320 --> 00:17:15.517 that produces the model called Kimmy, which is spelled Kim Mi, which is one of 00:17:15.600 --> 00:17:19.677 the leading open source models out of China. Um and uh the new version of 00:17:19.760 --> 00:17:23.197 Kimmy is a reasoning model that is at least according to the benchmark so far 00:17:23.280 --> 00:17:27.437 is basically a replication of the reasoning capabilities of GPT5, right? 00:17:27.520 --> 00:17:31.117 and and and these new models of GPT5 were a big advance over GPT4 and of 00:17:31.200 --> 00:17:34.477 course GPT5 costs a tremendous amount of money to to develop and to serve and all 00:17:34.560 --> 00:17:37.197 of a sudden you know here we are whatever 6 months later and you have an 00:17:37.280 --> 00:17:40.317 open source model called Kimmy and I think I don't know if they had it's 00:17:40.400 --> 00:17:42.957 either shrunk down to be able to run on either it's like one MacBook or two 00:17:43.040 --> 00:17:47.997 MacBooks um right um and so you can all of a sudden if you have like an applica 00:17:48.080 --> 00:17:50.877 you if you're a business and you want to have a reasoning model that's GPT5 00:17:50.960 --> 00:17:53.997 capable um but you you know you're whatever you're not going to pay the 00:17:54.080 --> 00:17:57.197 whatever GPT5 cost or you're not going to want to have it be hosted and you 00:17:57.280 --> 00:18:01.117 want to run it locally, um, you know, you can do that. Um, and and and again, 00:18:01.200 --> 00:18:04.237 that's just like another just it's just like another, you know, it's another 00:18:04.320 --> 00:18:06.797 breakthrough. Like it's just it's another another Tuesday, another huge 00:18:06.880 --> 00:18:09.117 advance. It's like, oh my god. And then of course, it's like, all right, well, 00:18:09.200 --> 00:18:11.917 what is OpenAI going to do? Well, obviously they're going to go to GPT6, 00:18:12.000 --> 00:18:15.837 right? Uh, and you know, right? And so there there's this kind of lattering 00:18:15.920 --> 00:18:18.477 that's happening where the entire industry is moving forward. Um, the big 00:18:18.560 --> 00:18:21.517 models are getting more capable. The small models are kind of chasing them. 00:18:21.600 --> 00:18:24.797 Um uh and then um and then the small models provide you know completely 00:18:24.880 --> 00:18:30.157 different way to deploy um you know at at at at very low price points. Um and 00:18:30.240 --> 00:18:32.877 so yeah I think and and you know we'll we'll see what happens. I mean there 00:18:32.960 --> 00:18:35.037 there are some very smart people in the industry who think that ultimately 00:18:35.120 --> 00:18:37.517 everything only runs in the big models because obviously the big models are 00:18:37.600 --> 00:18:40.077 always going to be the smartest and so therefore you're always you know you're 00:18:40.160 --> 00:18:41.997 always going to want the most intelligent thing because why would you 00:18:42.080 --> 00:18:45.037 ever want something that's not the most intelligent thing for any application. 00:18:45.120 --> 00:18:47.837 You know the counterargument is just there's a huge number of tasks that take 00:18:47.920 --> 00:18:50.877 place in the economy and in the world that don't require Einstein. you know, 00:18:50.960 --> 00:18:54.557 where, you know, where, you know, 120 IQ person is great. You don't need a, you 00:18:54.640 --> 00:18:58.317 know, 160 IQ, you know, PhD in, you know, string theory. You just like have 00:18:58.400 --> 00:19:02.157 somebody who's competent and capable and it's great. Um, and so, you know, I I, 00:19:02.240 --> 00:19:05.517 you know, and I we've talked about this before. I tend to think the AI industry 00:19:05.600 --> 00:19:08.157 is going to be structured a lot like the computer industry ended up getting 00:19:08.240 --> 00:19:11.117 structured, which is you're going to have a small handful of basically the 00:19:11.200 --> 00:19:14.077 equivalent of supercomputers, which are these like giant, you know, kind of we 00:19:14.160 --> 00:19:17.517 call god models that are, you know, running in these giant data centers. Um 00:19:17.600 --> 00:19:21.517 and then and then you know I I I I I'm not like convinced on this but my my 00:19:21.600 --> 00:19:24.077 kind of working assumption is what happens is then you have this cascade 00:19:24.160 --> 00:19:27.837 down of smaller models all ultimately all the way the very small models that 00:19:27.920 --> 00:19:31.277 run on embedded systems right run on run on individual chips inside every you 00:19:31.360 --> 00:19:35.357 know physical item in the world. Um and that you know the smartest models will 00:19:35.440 --> 00:19:38.397 always be at the top but the volume of models will actually be the smaller 00:19:38.480 --> 00:19:41.357 models that proliferate out and right that's what happened with microchips. uh 00:19:41.440 --> 00:19:44.157 it's what happened with computers which became microchips and then it's what 00:19:44.240 --> 00:19:46.717 happened with operating systems and with with a lot of everything else that we 00:19:46.800 --> 00:19:51.277 built in software. Um so you know I tend to think that's what will happen. Um 00:19:51.360 --> 00:19:55.917 just quickly on the chip side um again like chips you if you look at the entire 00:19:56.000 --> 00:20:00.877 history of the chip industry uh uh shortages become gluts um and you get 00:20:00.960 --> 00:20:04.637 just you know like anytime there's a giant profit pool in a in a new chip 00:20:04.720 --> 00:20:08.877 category um you know somebody has a lead for a while and kind of gets you know um 00:20:08.960 --> 00:20:12.317 let's say the the the profits appropriate to what we u what we call 00:20:12.400 --> 00:20:16.637 robust market share um but in time what happens right is that that draws 00:20:16.720 --> 00:20:19.597 competition and of course you know that that that's happening right now. So 00:20:19.680 --> 00:20:22.397 Nvidia's, you know, Nvidia is an absolutely fantastic company, fully 00:20:22.480 --> 00:20:24.797 deserves the position that they're in, fully deserves the profits that they're 00:20:24.880 --> 00:20:27.997 generating, but they're now so valuable, generating so many profits that it's the 00:20:28.080 --> 00:20:30.637 bat signal of all time to the rest of the chip industry to figure out how to 00:20:30.720 --> 00:20:32.557 advance the state-of-the-art AI chips. 00:20:32.640 --> 00:20:35.277 Um, and that's, by the way, and that's already happening, right? And so you've 00:20:35.360 --> 00:20:38.077 got other major companies like AMD coming at them, and then you've got 00:20:38.160 --> 00:20:41.197 really significantly, you've got the hyperscalers building their own chips. 00:20:41.280 --> 00:20:44.157 Um, and so, you know, a bunch of the big a bunch of those kind of big tech 00:20:44.240 --> 00:20:45.357 companies are building their own ships. 00:20:45.440 --> 00:20:48.397 Um, and of course then the Chinese are building their own ships as well. Um and 00:20:48.480 --> 00:20:52.397 so it's just it's like pretty likely in 5 years that that you know AI chips will 00:20:52.480 --> 00:20:56.317 be you know cheap and plentiful at least in comparison to the situation today. Uh 00:20:56.400 --> 00:20:59.277 which again I think will you know will tend to be extremely positive for the 00:20:59.360 --> 00:21:02.237 economics of of the kinds of companies that we invest in. 00:21:02.320 --> 00:21:04.237 >> Yep. And that startups are also starting 00:21:04.320 --> 00:21:06.797 to go after new chip design as well which is exciting. 00:21:06.880 --> 00:21:09.037 >> Yeah. Well, that's the other thing is yeah, you have these disruptive startups 00:21:09.120 --> 00:21:11.597 and actually that just for a moment on the chips, they were not really big 00:21:11.680 --> 00:21:14.077 investors in chips because it's kind of a big it's kind of a big company thing, 00:21:14.160 --> 00:21:18.157 but um it's a little bit of historical happen stance that AI is running on 00:21:18.240 --> 00:21:23.197 quote unquote GPUs um you know which GPU stands for graphical processing unit. So 00:21:23.280 --> 00:21:25.997 um and basically just for people who haven't tracked this there were 00:21:26.080 --> 00:21:28.957 basically two kinds of chips that made the personal computer happen. the 00:21:29.040 --> 00:21:33.757 so-called CPU central processing unit which classically was the Intel x86 x86 00:21:33.840 --> 00:21:36.317 chip that's kind of the brain of the computer and then there was this other 00:21:36.400 --> 00:21:39.437 kind of chip called the GPU or graphical processing unit that was the sort of 00:21:39.520 --> 00:21:44.077 second chip in every PC that does all the graphics um and you know and this is 00:21:44.160 --> 00:21:47.357 graphics you know 3D graphics for gaming or for CAD CAM or for you know anything 00:21:47.440 --> 00:21:50.877 else you know Photoshop or for anything that involves you know lots of visuals 00:21:50.960 --> 00:21:54.557 and so the the kind of canonical architecture for a personal computer was 00:21:54.640 --> 00:21:58.797 a CPU and a GPU by the way same thing for smartphones um but by the way. And 00:21:58.880 --> 00:22:01.597 over time, you know, these have kind of merged and so like a lot of CPUs now 00:22:01.680 --> 00:22:04.957 have GPU capability built in. Actually, a lot of GPUs now have CPU capability 00:22:05.040 --> 00:22:07.517 built in. So this, you know, this has gotten fuzzy over time, but like that 00:22:07.600 --> 00:22:10.557 that was like the classic breakdown. But the fact that that was the classic 00:22:10.640 --> 00:22:13.837 breakdown, you know, kind of meant that while Intel had a you know, monopoly for 00:22:13.920 --> 00:22:18.877 a long time on CPUs, um there was this other market of GPUs which Nvidia um you 00:22:18.960 --> 00:22:22.477 know basically fought the GPU wars for 30 years and and and came out the winner 00:22:22.560 --> 00:22:25.357 like what was the best company in the space. But it was like a hyper 00:22:25.440 --> 00:22:27.917 competitive market for graphics processors. it was actually not that 00:22:28.000 --> 00:22:31.517 high margin and it was actually not that big. And then basically it just it 00:22:31.600 --> 00:22:35.677 turned out that there were two other um forms of computation that were 00:22:35.760 --> 00:22:39.837 incredibly valuable that happened to be massively parallel uh in how they 00:22:39.920 --> 00:22:43.837 operate which which happened to be very good fits for the GPU architecture. And 00:22:43.920 --> 00:22:46.477 those two basically highly lucrative additional applications were 00:22:46.560 --> 00:22:50.717 cryptocurrency starting about you know 15 years ago and then AI starting about 00:22:50.800 --> 00:22:53.197 you know whatever four years ago. Um, 00:22:53.280 --> 00:22:58.317 and so and and Nvidia like I would say very cleverly set itself up with an 00:22:58.400 --> 00:23:01.437 architecture that works very well for this, but it's also just a little bit of 00:23:01.520 --> 00:23:04.637 a twist of fate that it just turns out that if AI is the killer app, it just 00:23:04.720 --> 00:23:08.237 turns out that the GPU architecture is the best legacy architecture is devoted 00:23:08.320 --> 00:23:11.517 to it. And I go through that to say like if you were designing AI chips from 00:23:11.600 --> 00:23:15.277 scratch today, you wouldn't build a full GPU. you would build dedicated AI chips 00:23:15.360 --> 00:23:19.277 that were much more much more specifically adapted to AI um and would 00:23:19.360 --> 00:23:22.717 have I I think would just be much more economically efficient and you know John 00:23:22.800 --> 00:23:25.277 to your point there there there are startups that are actually building 00:23:25.360 --> 00:23:29.917 entirely new kinds of chips uh oriented specifically for AI and you know we'll 00:23:30.000 --> 00:23:32.477 have to see what happens there you know it's hard to build a new chip company 00:23:32.560 --> 00:23:35.597 from scratch um you know it's possible that one or more of those startups makes 00:23:35.680 --> 00:23:39.357 it on their own um and some of them are you know doing very well um it's also 00:23:39.440 --> 00:23:42.797 possible of course that they get bought um you know by big companies that that 00:23:42.880 --> 00:23:46.797 have the ability to scale them. Um, and so, you know, you know, we'll see 00:23:46.880 --> 00:23:50.557 exactly how that unfolds. Um, and of course, we'll also, by the way, see, you 00:23:50.640 --> 00:23:54.397 know, the Koreans are going to play here for sure. Um, uh, the Japanese are going 00:23:54.480 --> 00:23:58.317 to play. Um, and then, you know, the Chinese in a major way, uh, as well. 00:23:58.400 --> 00:24:00.637 And, you know, they have their own, you know, native chip ecosystem that they're 00:24:00.720 --> 00:24:03.677 that they're building up. And so there there there there are going to be many 00:24:03.760 --> 00:24:06.877 choices of AI chips in the future. Um, and it's going to be that, you know, 00:24:06.960 --> 00:24:09.837 that'll be a giant battle that'll be a giant battle that we observe very 00:24:09.920 --> 00:24:13.517 carefully. um and that we uh make sure that our our companies basically are 00:24:13.600 --> 00:24:15.757 able to take full advantage of. 00:24:15.840 --> 00:24:19.757 >> While while on the topic of of international um we you mentioned Kimmy 00:24:19.840 --> 00:24:22.877 earlier. So it seems like some of the best open source models today are from 00:24:22.960 --> 00:24:27.677 China. Should this be worrisome to to folks? How are you thinking and talking 00:24:27.760 --> 00:24:29.757 about this topic with with folks in DC? 00:24:29.840 --> 00:24:31.517 I know you were just there last week. 00:24:31.600 --> 00:24:34.877 How much of this is a concern for uh US 00:24:34.960 --> 00:24:37.117 companies particularly just having seen 00:24:37.200 --> 00:24:40.477 the rise of China do unnatural things in 00:24:40.560 --> 00:24:45.757 solar markets, car markets? Um are they kind of flooding the ecosystem so that 00:24:45.840 --> 00:24:49.837 they can eventually kind of take share and and increasingly uh own the the 00:24:49.920 --> 00:24:52.957 ecosystem? >> Yeah. So uh you know a couple things. So 00:24:53.040 --> 00:24:56.077 one is you know you know you want to start these discussions by just kind of 00:24:56.160 --> 00:24:58.637 saying like you know look there's there's vigorous debate in in the US and 00:24:58.720 --> 00:25:01.197 around the world of look like you know how much are we in a new cold war with 00:25:01.280 --> 00:25:04.637 China you know and exactly like how hostile you know should should we view 00:25:04.720 --> 00:25:07.437 them and it you know it's very tempting by the way it's very tempting and I 00:25:07.520 --> 00:25:10.637 think it's a very good case made that we're in like a new cold war that's like 00:25:10.720 --> 00:25:14.397 that in a lot of ways is like the US versus USSR um in the in the 20th 00:25:14.480 --> 00:25:18.397 century um you know it is the counter argument would be it is more complicated 00:25:18.480 --> 00:25:21.517 than that because the US and the USSR were never really intertwined from a 00:25:21.600 --> 00:25:25.677 trade standpoint Um and and a big part of that quite frankly was the USSR never 00:25:25.760 --> 00:25:29.757 really made anything that anybody else needed I guess other than weapons. Um 00:25:29.840 --> 00:25:33.357 but like you know the USSR's primary exports were literally like you know 00:25:33.440 --> 00:25:36.077 literally like wheat and and oil. Um 00:25:36.160 --> 00:25:38.237 whereas of course China exports just a 00:25:38.320 --> 00:25:43.357 tremendous number of physical things right um including like a huge part of 00:25:43.440 --> 00:25:45.917 like the entire supply chain of parts that basically go into everything that 00:25:46.000 --> 00:25:50.157 American manufacturers you know kind of make right and so by the time a US you 00:25:50.240 --> 00:25:53.357 know whatever by the time an American company brings a toy to market right or 00:25:53.440 --> 00:25:57.677 a uh you know or a car um or anything or a computer or a smartphone or whatever 00:25:57.760 --> 00:26:00.877 like it's got a lot of componentry in it that was made in China so there so there 00:26:00.960 --> 00:26:04.317 is a much tighter in interlinkage between the the American and Chinese 00:26:04.400 --> 00:26:08.317 economies than there as the American and Soviet economies and you know may maybe 00:26:08.400 --> 00:26:10.877 you know Adam Smith or whatever might say you know that's good news for peace 00:26:10.960 --> 00:26:13.677 and that you know both countries need each other by the way the other part of 00:26:13.760 --> 00:26:16.477 that argument is that the Chinese basically the Chinese you know the 00:26:16.560 --> 00:26:20.477 Chinese governance model is based on high employment um you know because you 00:26:20.560 --> 00:26:23.517 know if if if you know at least all the geopolitical people say if China ended 00:26:23.600 --> 00:26:26.877 up with like 25 or 50% unemployment that would cause civil unrest which is the 00:26:26.960 --> 00:26:30.637 one thing that the CCP doesn't want and so the corresponding part of the trade 00:26:30.720 --> 00:26:33.997 pressure is China needs the American export market you know the American 00:26:34.080 --> 00:26:37.197 consumer is like a third of the global economy. Uh a third of global consumer 00:26:37.280 --> 00:26:41.757 demand. Um and so you know China needs the US export market or it has high all 00:26:41.840 --> 00:26:44.397 of a sudden a lot of its factories would go kind of instantly bankrupt and you 00:26:44.480 --> 00:26:47.437 know would cause mass unemployment and unrest in China. So so anyway like you 00:26:47.520 --> 00:26:51.517 know we there is this complicated it's a it's a complicated intertwined um 00:26:51.600 --> 00:26:55.917 relationship. Um having said that you know the the mood in DC basically for 00:26:56.000 --> 00:27:00.557 the last 10 years on a bipartisan basis um has been that we need to take we the 00:27:00.640 --> 00:27:04.637 US need to take China more seriously as a geopolitical foe. And you know under 00:27:04.720 --> 00:27:07.597 under under that school of thought there's sort of the sort of you know 00:27:07.680 --> 00:27:10.957 there's there's the military dimension which is you know the sort of the you 00:27:11.040 --> 00:27:13.997 know the the risk of some kind of war in the South China Sea the risk of some 00:27:14.080 --> 00:27:17.677 kind of war around around Taiwan and so that you know that that has everybody in 00:27:17.760 --> 00:27:21.357 Washington on high alert um you know there's also this this economic question 00:27:21.440 --> 00:27:24.877 around the kind of de-industrialization of the US potential re-industrialization 00:27:24.960 --> 00:27:28.157 and what that means about you know dependence on China and then and then 00:27:28.240 --> 00:27:31.997 there's and then there's this this this AI question um and and the AI question 00:27:32.080 --> 00:27:35.437 is an economic question but It's also like a geopolitical question which is 00:27:35.520 --> 00:27:38.637 okay you know basically AI is essentially only being built in the US 00:27:38.720 --> 00:27:43.037 and in China. Um you know the rest of the world either you know can't build it 00:27:43.120 --> 00:27:46.397 or doesn't want to which which we could talk about. So it's basically US versus 00:27:46.480 --> 00:27:50.237 China. Um and then AI is going to proliferate all over the world and is it 00:27:50.320 --> 00:27:52.557 going to be American AI that proliferates all over the world or is it 00:27:52.640 --> 00:27:55.677 going to be Chinese AI that proliferates all over the world and so and I was 00:27:55.760 --> 00:28:00.077 saying just generally across party lines in DC this you know the the things I 00:28:00.160 --> 00:28:04.797 just went through are kind of how they look at it. Um and and the Chinese are 00:28:04.880 --> 00:28:07.117 in the game and so the you know the Chinese are in the game for sure you 00:28:07.200 --> 00:28:10.237 know with software u you know deepseek you know was kind of the big you know 00:28:10.320 --> 00:28:13.357 kind of fire the starting gun in the software race and now you've got I think 00:28:13.440 --> 00:28:17.917 it's I think you've got four it's like deepsek uh which is a deep so deepseek 00:28:18.000 --> 00:28:22.477 is an AI model from actually a hedge fund um in uh in China um it's a little 00:28:22.560 --> 00:28:26.317 bit uh kind of took a lot of people by surprise um then Quen is the model from 00:28:26.400 --> 00:28:28.077 Alibaba. Kimmy is from another startup. 00:28:28.160 --> 00:28:32.157 Oh, called Moonshot. The company's called Moonshot. Um, and then there's, 00:28:32.240 --> 00:28:35.757 you know, and then, um, you know, there's also Tencent and BU. Um, and, 00:28:35.840 --> 00:28:38.797 um, by Dance, um, you know, that are all primary, you know, companies doing a lot 00:28:38.880 --> 00:28:42.077 of work in AI. Um, and so, you know, there's somewhere between three to six, 00:28:42.160 --> 00:28:43.597 you know, kind of primary AI companies. 00:28:43.680 --> 00:28:47.197 And then there's, you know, tremendous numbers of of startups. Um, and so, you 00:28:47.280 --> 00:28:50.157 know, they're in the race on on, uh, you know, they're in the race on on on 00:28:50.240 --> 00:28:53.757 software. Um, they are, you know, working to catch up on chips. They're 00:28:53.840 --> 00:28:56.557 not there yet, but they're working incredibly hard to catch up. And just as 00:28:56.640 --> 00:29:00.397 an example of that, you know, the at least the common understanding um you 00:29:00.480 --> 00:29:03.357 know, in the US is that the reason you haven't seen the new version of DeepSeek 00:29:03.440 --> 00:29:05.837 yet is that basically the Chinese government has instructed them to build 00:29:05.920 --> 00:29:09.597 it only on Chinese chips um as a as a motivator to get the Chinese chip 00:29:09.680 --> 00:29:12.717 ecosystem up and running. Um and and then the main chip company there is 00:29:12.800 --> 00:29:17.517 Huawei, although there could be more in the future. Um and then there's um so 00:29:17.600 --> 00:29:20.397 you know, so so so there's that and then and then there's everything to follow 00:29:20.480 --> 00:29:24.477 which is basically AI in kind of robotic form, right? And so there there's this 00:29:24.560 --> 00:29:28.957 basically global technological economic robotics competition that's kicking off. 00:29:29.040 --> 00:29:32.877 Um and u you know China kind of starts out ahead on robotics because they're 00:29:32.960 --> 00:29:36.957 just ahead on so many of the so many of the components that go into robots u 00:29:37.040 --> 00:29:39.757 because the you know the sort of like I said this the kind of entire supply 00:29:39.840 --> 00:29:43.197 chain of like electromechanical things you know basically moved from the US to 00:29:43.280 --> 00:29:47.117 China 30 years ago and and has never come back. So so so that's kind of the 00:29:47.200 --> 00:29:49.997 the the the DC lens on it. Um and and I 00:29:50.080 --> 00:29:54.877 would say you know DC is watching it uh you know quite carefully. Um uh the the 00:29:54.960 --> 00:29:58.397 the the big kind of supernova moment this year was the deepseek release. The 00:29:58.480 --> 00:30:01.917 deepseek release was surprising on a number of fronts. Um one was just how 00:30:02.000 --> 00:30:06.157 good it was and again along this line of it took the capability set that were 00:30:06.240 --> 00:30:07.597 running in large models in the cloud and 00:30:07.680 --> 00:30:10.637 kind of shrunk it um onto a um you know 00:30:10.720 --> 00:30:15.757 into into a uh into a a sort of a a reduced size you know a smaller version 00:30:15.840 --> 00:30:18.397 of sort of equivalent capabilities that you could run on small amounts of local 00:30:18.480 --> 00:30:22.157 hardware. Um and so there was that and then it was also a surprise that it was 00:30:22.240 --> 00:30:25.277 released as open source uh and particularly open source from China 00:30:25.360 --> 00:30:29.437 because China China does not have a long history of open source. Um and then um 00:30:29.520 --> 00:30:33.677 it was also a surprise um that it actually came from a hedge fund. Um so 00:30:33.760 --> 00:30:36.797 it didn't come from a big R&D you know sort of university research lab. It 00:30:36.880 --> 00:30:39.597 didn't come from a you know from a big tech company. it it came from a hedge 00:30:39.680 --> 00:30:43.997 fund and it it like as as far as we can tell it it basically is this somewhat 00:30:44.080 --> 00:30:47.197 idiosyncratic situation where you just have this incredibly successful quant 00:30:47.280 --> 00:30:50.477 hedge fund with all these you know super geniuses um and the the founder of that 00:30:50.560 --> 00:30:54.557 hedge fund you know basically decided to build AI um and you know at least 00:30:54.640 --> 00:30:57.197 external indications are this was a surprise to even even the Chinese 00:30:57.280 --> 00:31:00.477 government it's it's impossible to prove you know what the Chinese government was 00:31:00.560 --> 00:31:03.597 surprised by or not but you know there's at least the atmospherics are that this 00:31:03.680 --> 00:31:06.957 was not exactly planned this was not a national champion tech company at the 00:31:07.040 --> 00:31:09.997 time that Deepseek was released it was it sort of came out of left field which 00:31:10.080 --> 00:31:13.197 by the way is very encouraging for the field that it was possible for somebody 00:31:13.280 --> 00:31:16.077 to do that kind of who was unknown right because it kind of means that maybe you 00:31:16.160 --> 00:31:19.197 don't need all these you know super genius superstar researchers maybe 00:31:19.280 --> 00:31:22.157 actually smart kids can just build this stuff which I think is is the direction 00:31:22.240 --> 00:31:26.717 things are headed um and so that kicked off I would say like this kind of I I 00:31:26.800 --> 00:31:29.757 don't know copycat's the wrong word but that that was sort of it feels like the 00:31:29.840 --> 00:31:33.117 success of deepseek and the success of deepseek from China as open source kind 00:31:33.200 --> 00:31:37.437 of kicked off a sort of trend in China releasing these open source models um 00:31:37.520 --> 00:31:41.437 you know Look, the cynics, you know, in DC would say, you know, yeah, like 00:31:41.520 --> 00:31:44.477 they're dumping, right? The the they're obviously dumping. They're trying to, 00:31:44.560 --> 00:31:46.797 you know, they see that the West has this opportunity to build this China 00:31:46.880 --> 00:31:49.357 industry. You know, they're trying to commoditize it right out of the gate. 00:31:49.440 --> 00:31:52.797 You know, there's probably something to that. Um, you know, the the Chinese 00:31:52.880 --> 00:31:55.757 industrial economy does have a history of, you know, sort of, let's say, 00:31:55.840 --> 00:31:59.117 subsidized production that leads to selling, you know, selling things below 00:31:59.200 --> 00:32:04.237 cost in some cases. Um, but I think also it's it like I think that's almost too 00:32:04.320 --> 00:32:06.637 cynical of a view also because it's just like all right wow like they're really 00:32:06.720 --> 00:32:09.917 in the race like open source closed source whatever like that you know 00:32:10.000 --> 00:32:13.037 they're actually really in the race. Um, you know, we we've talked in the past, I 00:32:13.120 --> 00:32:16.477 think, on on on LP calls about, you know, these policy fights that, you 00:32:16.560 --> 00:32:18.717 know, we've been having in DC for the last two years. And, you know, there was 00:32:18.800 --> 00:32:22.157 a big pretty pretty big push within the US government, you know, two years ago 00:32:22.240 --> 00:32:25.597 to basically, you know, restrict, uh, you know, or outright ban, you know, a 00:32:25.680 --> 00:32:30.077 lot of AI. Um, and, you know, it's very easy for a country that is the only game 00:32:30.160 --> 00:32:31.517 in town to have those conversations. 00:32:31.600 --> 00:32:34.557 It's quite another thing if you're actually in a foot race with China. Um, 00:32:34.640 --> 00:32:37.197 and so I think actually the the the the 00:32:37.280 --> 00:32:41.597 policy landscape in DC has I would say has improved dramatically as a 00:32:41.680 --> 00:32:44.477 consequence of sort of an awareness now that this is actually a two- horse race, 00:32:44.560 --> 00:32:47.357 not a one-horse race. >> For sure. Yeah. Actually on on the point 00:32:47.440 --> 00:32:48.797 I'll I'll jump ahead here to policy and 00:32:48.880 --> 00:32:51.117 regulation just because it seems like uh 00:32:51.200 --> 00:32:54.797 the current stance on on 50 different 00:32:54.880 --> 00:32:57.277 set of AI laws by state seems like a 00:32:57.360 --> 00:33:02.397 catastrophic uh way to to put us effectively with a 00:33:02.480 --> 00:33:06.717 uh or one of our our hands tied behind our our back here in terms of the the AI 00:33:06.800 --> 00:33:08.797 race. What's a state of plan on that? 00:33:08.880 --> 00:33:12.477 Are folks recognizing that that would be catastrophic for progress and 00:33:12.560 --> 00:33:16.397 development? Where do most people at least stand on that topic today? 00:33:16.480 --> 00:33:17.997 Yeah. So it's a little bit complicated. 00:33:18.080 --> 00:33:21.197 So I'll rewind to say like two years ago I was very worried about like really 00:33:21.280 --> 00:33:24.637 ruinous federal federal legislation on AI and there was there was we you know 00:33:24.720 --> 00:33:27.117 we engaged you know kind of very heavily at that point which we've talked about 00:33:27.200 --> 00:33:29.997 in the past and I think the good news on that is I think the risk of that sitting 00:33:30.080 --> 00:33:34.237 here today is very low. Um I there's very little mood in DC on either side of 00:33:34.320 --> 00:33:38.157 the aisle uh to really you know essentially there's very little there's 00:33:38.240 --> 00:33:40.637 very little interest in doing anything that would prevent us from beating 00:33:40.720 --> 00:33:46.077 China. Um so so you know on the federal side things things are much better now. 00:33:46.160 --> 00:33:48.557 There there will there will be issues and there are tensions in the system but 00:33:48.640 --> 00:33:53.437 like things are looking looking pretty good. Um that has translated Jen to your 00:33:53.520 --> 00:33:55.837 point that's translated a lot of the attention to the states and basically 00:33:55.920 --> 00:33:59.197 what's happened is you know under our system of of federalism uh you know the 00:33:59.280 --> 00:34:02.717 states get to pass their own laws on a lot of things. Um and so uh yeah, 00:34:02.800 --> 00:34:05.597 basically you know a lot of you know and and you know with these things it's 00:34:05.680 --> 00:34:07.757 always a combination. A lot of well-meaning people are trying to figure 00:34:07.840 --> 00:34:09.757 out what to do at the state level and then of course there's a lot of 00:34:09.840 --> 00:34:13.437 opportunism where AI is just the hot topic. And so if you're a you know 00:34:13.520 --> 00:34:15.917 aggressive up and cominging state legislator or whatever in some state and 00:34:16.000 --> 00:34:18.077 you want to run for governor and then president you know you want to kind of 00:34:18.160 --> 00:34:21.437 attach yourself to the heat. Um and so there's like a political motivation to 00:34:21.520 --> 00:34:25.437 to do state level stuff. Um yeah and sitting here today like we're tracking 00:34:25.520 --> 00:34:29.757 on the order of,200 bills across the 50 states. And by the way, um, not just the 00:34:29.840 --> 00:34:33.197 blue states, also the red states. Um, and so, you know, I'm I've, you know, 00:34:33.280 --> 00:34:35.997 for the last like 5 years or whatever, I spent a lot of time complaining about, 00:34:36.080 --> 00:34:38.397 uh, you know, kind of what Democratic politicians are threatening to do to 00:34:38.480 --> 00:34:42.077 attack. There's also a lot of Republicans, like Republicans are not a 00:34:42.160 --> 00:34:44.877 block on this. And there are quite a few like local Republican officials in 00:34:44.960 --> 00:34:47.917 different states, um, that that also, I think, have, you know, let's say, you 00:34:48.000 --> 00:34:49.917 know, misinformed or ill-advised, um, 00:34:50.000 --> 00:34:55.677 views and are trying to put together, uh, put out bad bills. um you know it's 00:34:55.760 --> 00:34:58.077 a little bit weird that this is happening and that you know the federal 00:34:58.160 --> 00:35:01.677 government does have regulation of interstate commerce um and you know 00:35:01.760 --> 00:35:05.437 technology AI kind of by definition is interstate like you know there's there's 00:35:05.520 --> 00:35:07.197 no AI company that just operates in 00:35:07.280 --> 00:35:09.437 California or just operates in you know 00:35:09.520 --> 00:35:13.037 Colorado or Texas um you know AI of all 00:35:13.120 --> 00:35:16.797 technologies AI is obviously something this this sort of national in scope um 00:35:16.880 --> 00:35:19.197 you know it's sort it's sort of obvious that the federal government should be 00:35:19.280 --> 00:35:22.877 the regulator not not not the states um but but the federal government need 00:35:22.960 --> 00:35:24.877 needs to assert itself needs to step in. 00:35:24.960 --> 00:35:28.477 There there was actually an attempt to do that. There was a um there was an 00:35:28.560 --> 00:35:32.477 attempt to add a moratorium of state level AI regulation that basically would 00:35:32.560 --> 00:35:35.277 would reserve the right of the federal government to regulate AI and sort of 00:35:35.360 --> 00:35:38.317 prevent the states from moving forward with these bills. That was I think part 00:35:38.400 --> 00:35:41.917 of the negotiation for the quote one big beautiful bill and then that that there 00:35:42.000 --> 00:35:44.637 was a deal behind that and that deal kind of blew up at the at the last 00:35:44.720 --> 00:35:48.077 minute and that moratorium didn't happen and and you know in fairness the critics 00:35:48.160 --> 00:35:51.997 of that moratorum it probably was a was it was probably too much of a stretch. 00:35:52.080 --> 00:35:54.397 Well, it was I'm sorry. It was definitely too much of a stretch to get 00:35:54.480 --> 00:35:57.357 enough support to pass, but it was also probably too much of a stretch in terms 00:35:57.440 --> 00:35:59.757 of restricting the states from certain kinds of regulation that they really 00:35:59.840 --> 00:36:03.037 should be able to do. So, so it just it didn't quite come together. Um, there's 00:36:03.120 --> 00:36:06.317 a very active we're having very active discussions in DC right now about kind 00:36:06.400 --> 00:36:09.837 of the next, you know, the kind of the next turn on that. Um, you know, the 00:36:09.920 --> 00:36:12.317 administration is I would say the administration is very supportive of of 00:36:12.400 --> 00:36:15.197 the idea of of the federal government being in charge of this as part of it 00:36:15.280 --> 00:36:19.837 being an actual, you know, 50-st state issue. Um, and and and an issue of 00:36:19.920 --> 00:36:23.917 national importance. Um, and then, you know, I'd say most most Congress people 00:36:24.000 --> 00:36:27.357 on both sides of the aisle, you know, kind of get this. Um, so we just we we 00:36:27.440 --> 00:36:29.917 kind of have to figure out a way to, you know, to land this, but but I think 00:36:30.000 --> 00:36:32.397 that'll happen. Um, some of the state 00:36:32.480 --> 00:36:35.197 level bills are wild. Um, the the 00:36:35.280 --> 00:36:38.077 Colorado passed a very draconian uh 00:36:38.160 --> 00:36:41.357 regulation bill uh last year. Um, and 00:36:41.440 --> 00:36:45.357 against like furious objections from the local startup ecosystem in in in around 00:36:45.440 --> 00:36:48.557 Denver and Boulder. Um, and actually they're they're now actually trying to 00:36:48.640 --> 00:36:51.997 reverse their way out of that bill. um you know a year later some of the the 00:36:52.080 --> 00:36:55.277 nuance of it like the algorithmic discrimination and like how to mitigate 00:36:55.360 --> 00:36:59.197 like what were some of the the extreme versions of what they they had proposed. 00:36:59.280 --> 00:37:01.997 >> Yeah. So the really draconian one was the the one that we really fought hard 00:37:02.080 --> 00:37:05.757 was the one in California which was called SB1047 and it wasn't it it it was 00:37:05.840 --> 00:37:09.757 basically it was modeled basically after the was called the EU AI act. So the 00:37:09.840 --> 00:37:13.037 European Union's AI act. Okay. And this is the backdrop to all the US stuff 00:37:13.120 --> 00:37:16.317 which is the EU passed this bill called the AI act I don't know whatever two 00:37:16.400 --> 00:37:19.677 years ago and it basically has killed AI development in well it's actually killed 00:37:19.760 --> 00:37:24.317 AI development in Europe to a large extent. Um and then it even it it's so 00:37:24.400 --> 00:37:27.997 draconian that even even big American companies like Apple and Meta are not 00:37:28.080 --> 00:37:31.597 launching leading edge AI capabilities in their products in Europe. Like that 00:37:31.680 --> 00:37:34.317 that's how that's how like draconian that bill was. And it's it's sort of a 00:37:34.400 --> 00:37:38.557 classic it's a classic kind of European thing where they like you know like they 00:37:38.640 --> 00:37:41.357 just thought that you know they they have this kind of view that it's just 00:37:41.440 --> 00:37:43.917 like well you know we if we can't be the leader they literally say this by the 00:37:44.000 --> 00:37:45.917 way if we can't be the leaders in innovation at least we can be the 00:37:46.000 --> 00:37:48.477 leaders in regulation. Um and and and 00:37:48.560 --> 00:37:53.037 then they pass this like incredibly you know kind of ruinous uh selfharm you 00:37:53.120 --> 00:37:55.837 know kind of thing and then you know a few years pass and they're like oh my 00:37:55.920 --> 00:37:58.397 god what have we done and so they're you know they're kind of going through their 00:37:58.480 --> 00:38:02.957 own version of that. Um, by the way, you know, I I you know, when I talk about 00:38:03.040 --> 00:38:05.357 Europe, I I tend to be very dark about the whole thing. I will tell you the 00:38:05.440 --> 00:38:07.997 darkest people I know about Europe are the European entrepreneurs who moved to 00:38:08.080 --> 00:38:10.957 the US. Um, are just like absolutely 00:38:11.040 --> 00:38:15.197 furious about what's happening in in in Europe on this stuff. Um, but but even 00:38:15.280 --> 00:38:17.757 there, like it it's it's so bad in Europe, like they they shot themselves 00:38:17.840 --> 00:38:20.717 in the foot so badly that there's actually a process now at the at the EU 00:38:20.800 --> 00:38:22.077 to try to unwind that. They're trying to 00:38:22.160 --> 00:38:27.757 unwind the GDPR. So u anyway for people tracking Europe uh Mario Draghi um is 00:38:27.840 --> 00:38:30.397 the former I guess prime minister of Italy did this thing about a year ago 00:38:30.480 --> 00:38:33.677 called the Draghy report which is the report on European competitiveness and 00:38:33.760 --> 00:38:36.157 he kind of outlined kind of in great detail all the ways that Europe was 00:38:36.240 --> 00:38:39.277 holding itself back and part of it was overregulation areas like AI. So so 00:38:39.360 --> 00:38:43.277 they're trying to reverse out of that or making gestures you know we'll we'll see 00:38:43.360 --> 00:38:48.157 what happens. Um it in the middle of all that, California 00:38:48.240 --> 00:38:51.517 sort of inexplicably decided to basically copycat the EU AI act and try 00:38:51.600 --> 00:38:55.277 to apply it to California. Um which might strike you as completely insane. 00:38:55.360 --> 00:38:59.597 To which I would say yes, welcome to California. Um uh and um you know, it 00:38:59.680 --> 00:39:03.597 was this basically this like Sacramento political dynamic that kind of got got 00:39:03.680 --> 00:39:06.877 crazy. Um it would have you know completely killed you know AI 00:39:06.960 --> 00:39:10.877 development in California. Um unfortunately our governor vetoed it at 00:39:10.960 --> 00:39:14.237 the last minute. Um it did pass both houses legislature that he vetoed at the 00:39:14.320 --> 00:39:17.997 last minute. Um it to Jen to your point it would have done for it would have 00:39:18.080 --> 00:39:20.717 done a whole bunch of things that were ruinously uh bad. But one of the things 00:39:20.800 --> 00:39:21.757 it would have done is it would have 00:39:21.840 --> 00:39:24.717 assigned downstream liability um uh to 00:39:24.800 --> 00:39:28.957 open source developers. Um and so you know we talked about you know this 00:39:29.040 --> 00:39:31.277 Chinese open source thing. Okay so you got Chinese out there with open source. 00:39:31.360 --> 00:39:34.237 Now you're gonna have American companies that have open source AI. And by the way 00:39:34.320 --> 00:39:36.717 you're also going to have American academics and just like independent 00:39:36.800 --> 00:39:40.077 people in their nights and weekends developing open source. um you know 00:39:40.160 --> 00:39:43.117 which is a key way that all this technology proliferates and and so this 00:39:43.200 --> 00:39:46.477 this law would have assigned downstream liability to any misuse of open source 00:39:46.560 --> 00:39:49.997 to the original developer of the open source and so you know you're an 00:39:50.080 --> 00:39:53.197 independent developer or you're an academic or you're a startup you develop 00:39:53.280 --> 00:39:56.477 and release an AI model the AI model works fine the day you release it it's 00:39:56.560 --> 00:39:59.677 great but like 5 years later it gets built into a nuclear power plant and 00:39:59.760 --> 00:40:02.557 then there's a meltdown of the nuclear power plant and then somebody says oh 00:40:02.640 --> 00:40:05.357 it's the fault of the AI um the the the 00:40:05.440 --> 00:40:09.757 the legal liability for that nuclear meltdown or for anything any other 00:40:09.840 --> 00:40:12.557 practical real world thing that would follow in the out years would then be 00:40:12.640 --> 00:40:15.277 assigned back to that open source developer. Of course, this is completely 00:40:15.360 --> 00:40:18.557 insane. It would completely kill open source. It would completely kill 00:40:18.640 --> 00:40:21.357 startups doing open source. It would completely kill academic research like 00:40:21.440 --> 00:40:25.677 in its entirety. Um, you know, anything in the field. Um, and so, you know, that 00:40:25.760 --> 00:40:28.957 like that's the level of playing with fire. Um, you know, kind of that these 00:40:29.040 --> 00:40:32.557 state level politicians have become enamored with. Um, like I said, I think 00:40:32.640 --> 00:40:35.117 the good news is the feds understand this. I suspect that this is going to 00:40:35.200 --> 00:40:38.557 get resolved, but it but it does need to get resolved because, you know, just as 00:40:38.640 --> 00:40:40.957 a country, it just doesn't make any sense to let let the states kind of 00:40:41.040 --> 00:40:46.077 operate suicidally like this. Um, and so that's what we're doing. You know, we we 00:40:46.160 --> 00:40:49.277 talk about this, we call this our little tech agenda. Um, we're extremely focused 00:40:49.360 --> 00:40:52.957 on on on the freedom and starters innovate. We are not trying to argue, 00:40:53.040 --> 00:40:56.557 you know, many many other issues. We operate in a completely bipartisan 00:40:56.640 --> 00:40:59.997 fashion. We have extensive um support, you know, on both sides of the aisle and 00:41:00.080 --> 00:41:04.397 for both sides of the aisle. Um, so it's it's a truly bipartisan effort. um very 00:41:04.480 --> 00:41:07.277 policy based and you know I think very much aligned with the interests of the 00:41:07.360 --> 00:41:12.077 country uh broadly um and so that is what we're doing and then and then the 00:41:12.160 --> 00:41:15.357 other question we get we we get actually you know in some cases from LP but in a 00:41:15.440 --> 00:41:17.357 lot of cases actually from employees um 00:41:17.440 --> 00:41:22.557 is like okay why us right like you know you know with with any sort of you know 00:41:22.640 --> 00:41:25.117 policy question like this there's always this collective action question which is 00:41:25.200 --> 00:41:28.237 just like you know tragedy of the commons which is in theory like 00:41:28.320 --> 00:41:31.117 everybody every venture firm every tech company whatever should be weighing in 00:41:31.200 --> 00:41:34.877 on these things in practice what happens is mo most them just simply don't. Um, 00:41:34.960 --> 00:41:37.437 and so at some point it falls on somebody's shoulders to fight these 00:41:37.520 --> 00:41:41.117 things. And we we we Ben and I just basically concluded that the stakes here 00:41:41.200 --> 00:41:44.157 were just way too high. You know, if if we're going to be the industry leader, 00:41:44.240 --> 00:41:48.237 we just have to take responsibility for our own destiny. You know, for better or 00:41:48.320 --> 00:41:50.797 for worse, I think that's the cost of doing business uh for being the leader 00:41:50.880 --> 00:41:53.597 in the field right now. >> Before we get off the topic of of AI, I 00:41:53.680 --> 00:41:56.477 want to go back to one question that that was submitted in. So, do you think 00:41:56.560 --> 00:42:01.677 usage based or utility is a right way to price an AI compared to seeds? 00:42:01.760 --> 00:42:05.117 Ah that is a fantastic question. So this is one of these giant this is in my my 00:42:05.200 --> 00:42:08.157 list of what I call the trillion dollar questions u where you know depending on 00:42:08.240 --> 00:42:11.357 how this is answered will drive you know trillions of dollars of market value. So 00:42:11.440 --> 00:42:15.037 yeah so usage based pricing it's it's actually 00:42:15.120 --> 00:42:17.437 it's actually fairly amazing if you think about this from a startup 00:42:17.520 --> 00:42:20.397 standpoint from a venture standpoint it's actually fairly amazing what's 00:42:20.480 --> 00:42:22.957 happened and I'm trying I'm not really talking about this in public because I 00:42:23.040 --> 00:42:25.597 don't really I because I don't want it to stop. I think it's actually quite 00:42:25.680 --> 00:42:29.997 amazing. Um, which is you have these technology companies, you know, these 00:42:30.080 --> 00:42:32.797 big tech companies with these like incredible R&D capabilities that are 00:42:32.880 --> 00:42:35.677 building these big models, these big AI models with this incredible, you know, 00:42:35.760 --> 00:42:39.357 new new kind of new new kind of intelligence. And then it it turns out 00:42:39.440 --> 00:42:42.637 that they were already in a war. They were already in the cloud war, right? 00:42:42.720 --> 00:42:45.357 And so they were already in the war for kind of cloud services. And this is like 00:42:45.440 --> 00:42:48.797 AWS versus Azure versus uh Google Cloud. 00:42:48.880 --> 00:42:51.837 Um, you know, and then all the all these other all these other cloud efforts. And 00:42:51.920 --> 00:42:55.517 so what what what what actually happened was they sort of like there's an 00:42:55.600 --> 00:42:58.877 alternate universe in which they basically just kept all of their magic 00:42:58.960 --> 00:43:03.197 AI secret and captive and just used it in their own business um or used it to 00:43:03.280 --> 00:43:06.797 just compete with more companies um you know in more in more categories but 00:43:06.880 --> 00:43:10.397 instead what they've done is they've basically you know if I commod 00:43:10.480 --> 00:43:13.757 commoditize is too strong a word but they they have they have proliferated 00:43:13.840 --> 00:43:17.837 their magic new technology through their cloud business um which is which is this 00:43:17.920 --> 00:43:20.557 business that just has these like incredible scale you know kind of kind 00:43:20.640 --> 00:43:23.757 of components to But um you know and sort of this hyper competition between 00:43:23.840 --> 00:43:26.797 the providers and these you know these these prices that that come down very 00:43:26.880 --> 00:43:29.837 fast. Um, and so you've got like the most magic new technology in the world 00:43:29.920 --> 00:43:34.157 and then it's basically being served up by those companies in in in a as a cloud 00:43:34.240 --> 00:43:37.037 business and made made basically available to everybody on the planet to 00:43:37.120 --> 00:43:40.797 just click and use and for like relatively small amounts of money and 00:43:40.880 --> 00:43:44.077 then on on a usage basis which means and usage is great for startups because you 00:43:44.160 --> 00:43:47.197 it means you can start easily right you the the the you know there's very you 00:43:47.280 --> 00:43:50.477 know there's basically no fixed co for a startup building an AI app they don't 00:43:50.560 --> 00:43:53.597 have giant fixed cost because they could just tap into the open AI or anthropic 00:43:53.680 --> 00:43:57.117 or Google or Microsoft or whatever you know cloud you know tokens by the you 00:43:57.200 --> 00:44:00.397 know, intelligence tokens by the drink offering and just get going. Um, and so 00:44:00.480 --> 00:44:03.517 it's it's kind of this this from this from the startup standpoint, it's like 00:44:03.600 --> 00:44:06.077 this marvelous thing where like the most magical thing in the world is available 00:44:06.160 --> 00:44:11.357 by the drink. You know, it's absolutely amazing. Um, uh, I, you know, and, you 00:44:11.440 --> 00:44:13.517 know, that model, you know, by the way, that model's working and those companies 00:44:13.600 --> 00:44:15.757 are happy and they're growing really fast and they're, you know, happily 00:44:15.840 --> 00:44:18.557 reporting massive cloud revenue growth and, you know, they they're happy with 00:44:18.640 --> 00:44:21.837 the margins and so forth and so, you know, I think generally it's working. 00:44:21.920 --> 00:44:25.357 Um, and those businesses are, I think, likely to get much larger. Um and so I 00:44:25.440 --> 00:44:28.637 think you know generally that's going to work but but to to to the question like 00:44:28.720 --> 00:44:32.397 that doesn't mean that the optimal pricing model for for example all of the 00:44:32.480 --> 00:44:35.757 applications should be tokens by the drink and in fact very much I think not 00:44:35.840 --> 00:44:39.357 the case. Um you know we spend a lot of time working we actually have you know 00:44:39.440 --> 00:44:42.797 dedicated you know experts on on pricing in our firm. We spend a lot of time with 00:44:42.880 --> 00:44:45.757 our companies working on pricing because it's you know it's really this magical 00:44:45.840 --> 00:44:48.077 art and science that that a lot of companies don't take don't take 00:44:48.160 --> 00:44:50.797 seriously enough. So we spend a lot of time with other companies on this. And 00:44:50.880 --> 00:44:54.397 of course, you know, a core principle of pricing is you don't want to price by 00:44:54.480 --> 00:44:58.317 cost if you can avoid it. You want to price by value, right? Like you want to 00:44:58.400 --> 00:45:01.757 price you price where you're getting a percentage of the business value um of, 00:45:01.840 --> 00:45:04.397 you know, especially when you're selling two businesses, you want to price as a 00:45:04.480 --> 00:45:07.837 percentage of the business value that you're getting. And so so you do have 00:45:07.920 --> 00:45:10.557 some AI startups that are that are pricing by the drink for certain things 00:45:10.640 --> 00:45:13.517 that they're doing, but you have many others that are exploring other pricing 00:45:13.600 --> 00:45:17.037 models. uh you know some that are just like replications of SAS pricing models 00:45:17.120 --> 00:45:19.677 but you also have other companies are explor exploring pricing models for 00:45:19.760 --> 00:45:22.077 example of well if the AI can actually 00:45:22.160 --> 00:45:24.717 do the job of a coder or the AI could do 00:45:24.800 --> 00:45:26.957 the job of a doctor or a nurse or a 00:45:27.040 --> 00:45:29.677 radiologist or a lawyer or a parallegal 00:45:29.760 --> 00:45:34.157 right or whatever or a teacher. Um you know basically can you can could can you 00:45:34.240 --> 00:45:37.517 price by value and can you get a percentage of the value of what of what 00:45:37.600 --> 00:45:40.397 of of of what otherwise would would would have been you know would have been 00:45:40.480 --> 00:45:44.237 literally a person. um you know or or by the way equivalently can you price by 00:45:44.320 --> 00:45:47.197 marginal productivity. So if you can take a human doctor and make them much 00:45:47.280 --> 00:45:49.917 more productive because you give them AI, you know, can you price as a 00:45:50.000 --> 00:45:53.117 percentage of kind of the productivity uplift, uh, you know, from the from from 00:45:53.200 --> 00:45:56.557 the from the augment, you know, the comb symbiotic relationship between the the 00:45:56.640 --> 00:46:00.797 human being and and the AI. Um, and so I I think what we see in startup land is 00:46:00.880 --> 00:46:04.397 like a lot of experimentation happening on on these pricing models. And I and I 00:46:04.480 --> 00:46:08.877 and I think again I I think that's like super healthy. Um, I I you know, I was 00:46:08.960 --> 00:46:11.517 in this little speech on this is like high prices are really underappreciated. 00:46:11.600 --> 00:46:14.877 High prices are often a favorite of the customer. It's actually really funny. A 00:46:14.960 --> 00:46:17.597 lot of like the naive view on pricing is the lower the price, the better it is 00:46:17.680 --> 00:46:19.997 for the customer. The the more sophisticated looking at it is higher 00:46:20.080 --> 00:46:22.637 prices are often good for the customer because a higher price means that the 00:46:22.720 --> 00:46:26.957 vendor can make the product better faster, right? Like you can actually 00:46:27.040 --> 00:46:29.917 companies with higher prices, higher margins can actually invest more in R&D 00:46:30.000 --> 00:46:33.277 and they can actually make the product better. Um and you know most people who 00:46:33.360 --> 00:46:35.677 buy things aren't just looking for the cheapest price. They want something 00:46:35.760 --> 00:46:38.957 that's really that's going to work really well. Um and so often high 00:46:39.040 --> 00:46:42.957 prices, you know, the customer doesn't ever say this. it'll never show up in a 00:46:43.040 --> 00:46:46.397 survey. Um, but but the high price can actually be a gift for the customer 00:46:46.480 --> 00:46:48.557 because it can make the vendor better, can make the product better, and 00:46:48.640 --> 00:46:49.757 ultimately make the customer better off. 00:46:49.840 --> 00:46:53.677 And so I I'm I'm very encouraged by the degree to which the AI entrepreneurs are 00:46:53.760 --> 00:46:55.997 willing to run these experiments. And I, you know, we'll have to see where it 00:46:56.080 --> 00:46:59.197 pans out. But at least so far, I feel I feel good about the the uh, you know, at 00:46:59.280 --> 00:47:00.957 least the attitude of the industry about it. 00:47:01.040 --> 00:47:04.077 >> Awesome. I actually uh I was, as you were gone through, I had probably 10 00:47:04.160 --> 00:47:07.437 more follow-up questions, but I'm actually going to go back to um a topic 00:47:07.520 --> 00:47:09.837 you had uh briefly, the trillion dollar 00:47:09.920 --> 00:47:14.557 questions. Will open source or close source win? Feels like we we've come out 00:47:14.640 --> 00:47:17.677 on this this debate or where do you where do you put that? 00:47:17.760 --> 00:47:20.957 >> No, I think this is still open. I I think this is still very open. Um you 00:47:21.040 --> 00:47:25.517 know that like the the the closed source models keep getting better. Um uh by the 00:47:25.600 --> 00:47:28.557 way if you generally if you just like take the temperature of the people 00:47:28.640 --> 00:47:31.677 working at the big labs who work on the big proprietary models like generally 00:47:31.760 --> 00:47:35.437 what they'll tell you is progress is continuing at a very rapid pace. Um you 00:47:35.520 --> 00:47:37.597 know there's there's this you know there's this periodic concern that kind 00:47:37.680 --> 00:47:40.557 of shows up on online which is or in the in the market which is like you know 00:47:40.640 --> 00:47:43.757 maybe the capabilities these models are topping out um and you know there's 00:47:43.840 --> 00:47:46.397 certain there's there's certain areas in which you know there's there's you know 00:47:46.480 --> 00:47:49.757 people are working but like the people working at the big labs are like oh no 00:47:49.840 --> 00:47:52.797 we have like 800 new idea like we have tons of new ideas we have tons of new 00:47:52.880 --> 00:47:56.397 ways of doing things. We we might need to find new ways to scale but like we we 00:47:56.480 --> 00:47:57.997 have a lot of ideas on how to do that. 00:47:58.080 --> 00:48:00.157 We know a lot of ways to make these things better and you know we're 00:48:00.240 --> 00:48:02.957 basically making new discoveries all the time. So like I would say you know 00:48:03.040 --> 00:48:06.237 generally the people working in the like across all the big labs are are pretty 00:48:06.320 --> 00:48:10.077 optimistic. Um and so like I I think the big models are going to continue to get 00:48:10.160 --> 00:48:13.597 better you know very quickly here and then you know overall um and then the 00:48:13.680 --> 00:48:16.717 open source models continue to get better. Um and like I said you know you 00:48:16.800 --> 00:48:19.117 know every every every I don't know every month or something there's like 00:48:19.200 --> 00:48:22.157 another big release of like something like this Kimmy thing. Um where it's 00:48:22.240 --> 00:48:24.877 just like wow like you know that's amazing and you know wow they really 00:48:24.960 --> 00:48:27.517 like shrunk that down and got that capability on a very small form factor. 00:48:27.600 --> 00:48:30.957 Um uh and so um yeah that's the case and 00:48:31.040 --> 00:48:34.557 then you know I maybe just the third kind of thing to bring up is um the 00:48:34.640 --> 00:48:38.477 other really nice benefit of open source um is that uh open source is the thing 00:48:38.560 --> 00:48:43.437 that's easy to learn from right um and so if you're a you know computer sc if 00:48:43.520 --> 00:48:47.437 you're a computer science professor who wants to teach a class on on CS on AI or 00:48:47.520 --> 00:48:49.917 if you're a computer science student that's trying to learn about it or if 00:48:50.000 --> 00:48:53.837 you're just like a normal engineer in a normal company trying to learn this new 00:48:53.920 --> 00:48:57.757 thing um or just somebody in your you know by the way somebody in basement at 00:48:57.840 --> 00:49:00.797 night with a startup idea. Um the existence of these of these 00:49:00.880 --> 00:49:03.837 state-of-the-art open source models is amazing because that's the education 00:49:03.920 --> 00:49:06.797 that you need. Like they actually these open source models actually show you how 00:49:06.880 --> 00:49:11.837 to do everything. Um right. Um and so like and and what that's leading to 00:49:11.920 --> 00:49:14.717 right is the proliferation of the knowledge about how to build AI is like 00:49:14.800 --> 00:49:18.557 expanding very fast. Um again as compared to a counterfactual world in 00:49:18.640 --> 00:49:22.077 which it was all basically bottled up in two or three big companies. And so, you 00:49:22.160 --> 00:49:25.277 know, the open source thing is also just proliferating knowledge and then that 00:49:25.360 --> 00:49:29.037 knowledge is generating a lot of new people. Um, and so I I you know, you 00:49:29.120 --> 00:49:32.397 know, as you guys have all seen sitting here today, AI researchers are at an 00:49:32.480 --> 00:49:35.357 enormous premium. You know, AI researchers today are getting paid more 00:49:35.440 --> 00:49:37.277 than professional athletes. Um, right? 00:49:37.360 --> 00:49:40.797 Like, you know, and that's right, that's a supply demand imbalance there. There 00:49:40.880 --> 00:49:45.197 aren't enough of them to go around. But, you know, again, shortages create glut. 00:49:45.280 --> 00:49:48.957 um the the number of the number of smart people in the world who are coming up to 00:49:49.040 --> 00:49:52.397 speed very quickly on how to build these things u I mean some of the best AI 00:49:52.480 --> 00:49:55.917 people in the world are like 22 23 24 like they you know kind of by definition 00:49:56.000 --> 00:49:58.957 they haven't been in the field that long you know you know they they can't have 00:49:59.040 --> 00:50:02.317 been experts their whole lives right so you know they they kind of have to have 00:50:02.400 --> 00:50:05.197 come up to speed over the course of the last four or five years and and if if 00:50:05.280 --> 00:50:07.757 they if they've been able to do that then then there's going to be a lot more 00:50:07.840 --> 00:50:11.277 in the future that are going to do that um and so just the the the sort of 00:50:11.360 --> 00:50:14.637 spread of the level of expertise on this technology is happening now very quickly 00:50:14.720 --> 00:50:18.077 Um, so I yeah, I mean I think it's still like I said, I think it's I think it's 00:50:18.160 --> 00:50:20.877 still a race. And and by the way, you know, look, the long-term answer may 00:50:20.960 --> 00:50:24.957 well just be both. Um, you know, like I said, if you if you believe my pyramid 00:50:25.040 --> 00:50:28.397 industry structure, then there will then there will certainly be a large business 00:50:28.480 --> 00:50:32.477 of whatever is the smartest thing almost regardless of how of how much it costs. 00:50:32.560 --> 00:50:35.917 Um, and then there but there will also be this just giant volume market of of 00:50:36.000 --> 00:50:38.717 smaller models everywhere, which which is what we're also seeing. 00:50:38.800 --> 00:50:42.557 >> Yep. Yep. The another question you had posed at at that point in time was will 00:50:42.640 --> 00:50:45.837 incumbents versus startups went and at that point in time I think there was a 00:50:45.920 --> 00:50:50.077 mixed bag of where the incumbents were approaching AI. I think that's radically 00:50:50.160 --> 00:50:52.157 changed in the last two years. Um and 00:50:52.240 --> 00:50:54.477 then on the counter example the the 00:50:54.560 --> 00:50:58.077 blossoming of startups increasingly now 00:50:58.160 --> 00:51:01.677 maybe migrating into the incumbent category just how big they since that 00:51:01.760 --> 00:51:05.357 time. You you want to take that uh question and and give uh your assessment 00:51:05.440 --> 00:51:07.757 of where where the state of the world is? 00:51:07.840 --> 00:51:09.917 >> Yeah. Yeah. So, I mean, look, you know, big companies that are definitely, you 00:51:10.000 --> 00:51:12.957 know, playing hard. You know, Google's playing hard. Meta's playing hard. Um, 00:51:13.040 --> 00:51:16.557 Amazon, um, Microsoft, um, you know, there's a bunch of these companies that 00:51:16.640 --> 00:51:19.517 are, you know, that are kind of in in in there, um, you know, very aggressively. 00:51:19.600 --> 00:51:21.677 And then you've got these, you know, what we call the new incumbents like 00:51:21.760 --> 00:51:25.917 Anthropic and and, uh, and Open AI. Um, but you also have like, you know, even 00:51:26.000 --> 00:51:28.557 in the last two years, you've had this birth of all of a sudden like brand new 00:51:28.640 --> 00:51:31.517 companies that are almost instant incumbents. And you, you could say XAI 00:51:31.600 --> 00:51:36.477 is one of those. Uh, ML, by the way, ML is the great outlier to my Europe thing 00:51:36.560 --> 00:51:39.837 from earlier. like Mald is actually doing very well as sort of the European 00:51:39.920 --> 00:51:43.677 kind of uh you know French national European uh continental you know kind of 00:51:43.760 --> 00:51:48.237 AI champion um sort of the you know the exception that proves the rule um but 00:51:48.320 --> 00:51:50.317 you know there's there's a bunch of these now that are like you know doing 00:51:50.400 --> 00:51:53.357 quite well and are kind of becoming new incumbents um and then of course there's 00:51:53.440 --> 00:51:56.077 tons of startups by the way there's and then there's there's actual foundation 00:51:56.160 --> 00:52:00.077 model startups right and so you know we funded uh you know we funded Ilas out of 00:52:00.160 --> 00:52:03.597 open AAI to do a new foundation model company we funded Miriam Maratti also 00:52:03.680 --> 00:52:06.877 out of open AI we funded Faith Ali out of Stanford to do a world 00:52:06.960 --> 00:52:09.837 foundation model company and so you know there you know there's there are new 00:52:09.920 --> 00:52:14.077 swings all all you know all early but very promising um for to kind of build 00:52:14.160 --> 00:52:17.837 you know new incumbents quickly um and so you know that's all happening and 00:52:17.920 --> 00:52:20.317 then and then you know what and then on top of that there's just this giant 00:52:20.400 --> 00:52:23.277 explosion of AI application companies right and so there there's basically 00:52:23.360 --> 00:52:27.197 companies that then usually startups that basically take the technology and 00:52:27.280 --> 00:52:30.877 then you know field it in a specific domain whether that's law or medicine or 00:52:30.960 --> 00:52:33.997 education or you know creativity um or 00:52:34.080 --> 00:52:36.957 or or or whatever Um but again here it's 00:52:37.040 --> 00:52:39.117 just like it's amazing kind of how how 00:52:39.200 --> 00:52:43.997 sophisticated things are getting very quickly. So 00:52:44.080 --> 00:52:46.717 talk about the application companies for a moment. So like an application company 00:52:46.800 --> 00:52:49.437 like classic example is like a cursor is like an application company. So they 00:52:49.520 --> 00:52:53.517 take the core AI capability which they purchase by the drink from you know 00:52:53.600 --> 00:52:57.037 anthropic or open AI or Google um you know to tokens by the drink and then 00:52:57.120 --> 00:53:01.997 they they they build a code basically a code editor what we used to call an IDE 00:53:02.080 --> 00:53:05.357 um integrated development environment or basically like a a software creation 00:53:05.440 --> 00:53:09.917 system um so they build like an AI coding system um on on top of the 00:53:10.000 --> 00:53:13.037 anthropic or open AAI or whatever you know kind of kind of big models feel 00:53:13.120 --> 00:53:16.077 that and that the the critique of those companies in the industry has been oh 00:53:16.160 --> 00:53:19.437 those are what are called called GPT rappers is kind of the pjorative And the 00:53:19.520 --> 00:53:22.557 idea basically being is well they're not actually like they're not actually doing 00:53:22.640 --> 00:53:26.237 anything that's going to preserve value because the the actual the the whole 00:53:26.320 --> 00:53:29.117 point of what they're doing is they're surfacing AI but it's not their AI. The 00:53:29.200 --> 00:53:32.157 the AI that's being surfaced is from somebody else. And so these are kind of 00:53:32.240 --> 00:53:35.357 these pass pass through shell things that ultimately won't have value. It 00:53:35.440 --> 00:53:38.077 actually turns out what's happening is kind of the opposite of that which is 00:53:38.160 --> 00:53:42.397 the the leading uh AI application companies like Cursor I mean f first of 00:53:42.480 --> 00:53:45.117 all what they're discovering is they they're not just using a single AI 00:53:45.200 --> 00:53:48.237 model. they're actually they actually as these products get more sophisticated 00:53:48.320 --> 00:53:50.877 they actually end up using many different kinds of models that are kind 00:53:50.960 --> 00:53:54.957 of customtailored to the specific aspects of how these products work. Um 00:53:55.040 --> 00:53:57.437 and so they may start out using one model but they end up using a dozen 00:53:57.520 --> 00:54:00.157 models and then in the fullness of time it might be 50 or 100 different models 00:54:00.240 --> 00:54:03.677 for different aspects of the product. A and then B they end up building a lot of 00:54:03.760 --> 00:54:07.517 their own models. Um and so they they a lot of these the leading edge 00:54:07.600 --> 00:54:09.837 application companies are actually backward integrating and actually 00:54:09.920 --> 00:54:12.717 building their own AI models because because they have the deepest 00:54:12.800 --> 00:54:14.957 understanding of their domain. and they're able to build the model that's 00:54:15.040 --> 00:54:19.597 best suited to that. Um, and then by the way, also AI open source, they're also 00:54:19.680 --> 00:54:23.437 able to pick up and run an open source models. Um, and so if they don't like 00:54:23.520 --> 00:54:27.437 the economics of of buying intelligence, you know, by the drink from a from a 00:54:27.520 --> 00:54:30.157 from a cloud service provider, you know, they can pick up one of these open 00:54:30.240 --> 00:54:32.477 source models and implement it instead, which, you know, which these companies 00:54:32.560 --> 00:54:36.397 are also doing. Um, and so the the best of the best of the AI application 00:54:36.480 --> 00:54:39.757 companies are they are actually full-fledged deep technology companies 00:54:39.840 --> 00:54:43.357 actually building their own AI. Um and so that you know that's I think 00:54:43.440 --> 00:54:46.397 >> small models though right Mark when you think about god models versus small 00:54:46.480 --> 00:54:49.117 models as you were describing that but that would be small would you categorize 00:54:49.200 --> 00:54:51.997 that as a small >> well some of them I mean we I will let 00:54:52.080 --> 00:54:54.477 them I will let them announce you know whatever they're doing whenever it's 00:54:54.560 --> 00:54:58.077 appropriate but some of them are now also doing big model development um and 00:54:58.160 --> 00:55:00.797 again this this is also part of what this is also part of the learning just 00:55:00.880 --> 00:55:04.157 in the last two years well so like here's a big learning just from the last 00:55:04.240 --> 00:55:06.477 two years which is very interesting which is two years ago or three years 00:55:06.560 --> 00:55:10.157 ago for sure you would have said wow open AI is like way out ahead um and 00:55:10.240 --> 00:55:12.237 like it's probably going to be impossible for anybody to catch up and 00:55:12.320 --> 00:55:14.717 then it's like okay well Anthropic caught up and so but you know they came 00:55:14.800 --> 00:55:17.677 out of open AI and so they had all the secrets you know whatever and so knew 00:55:17.760 --> 00:55:20.477 how to do it and so okay they caught up but surely nobody can catch up after 00:55:20.560 --> 00:55:23.837 them and then very quickly after that there were a raft of other companies 00:55:23.920 --> 00:55:27.357 that caught up very fast and and XAI is maybe the best example of that which is 00:55:27.440 --> 00:55:32.157 like you know XAI you know Elon's company XAI is the company name gro is 00:55:32.240 --> 00:55:35.917 the consumer product version of it um XAI basically caught up to you know 00:55:36.000 --> 00:55:39.597 state-of-the-art openai anthropic level in in like less than 12 months from a 00:55:39.680 --> 00:55:43.757 standing start right and So, and again that that kind of argues against any 00:55:43.840 --> 00:55:47.037 kind of permanent lead, right, by by any one incumbent that's just going to 00:55:47.120 --> 00:55:49.837 basically be able to lock the entire market down like if you can catch up 00:55:49.920 --> 00:55:52.877 like that. And then and then as we as we've discussed the you know the China 00:55:52.960 --> 00:55:54.797 part is all new in the last year, right? 00:55:54.880 --> 00:55:58.877 The deepseek uh this the deepseek moment I think was in January or February of 00:55:58.960 --> 00:56:02.877 this year, right? So less than 12 months ago. Um and so and now you've got like 00:56:02.960 --> 00:56:06.077 four Chinese companies that have effectively caught up. And so, you know, 00:56:06.160 --> 00:56:09.277 so it's like, all right, I mean, again, this is these are these are trillion 00:56:09.360 --> 00:56:13.037 dollar questions, not answers. But it's just like, wow, okay, like it's one of 00:56:13.120 --> 00:56:16.717 these things where once somebody proves that it's capable, it seems to not be 00:56:16.800 --> 00:56:19.837 that hard for other people to be able to catch up, even people with far less 00:56:19.920 --> 00:56:23.517 resources. Um, and so, you know, I don't know what that does. Maybe it makes you 00:56:23.600 --> 00:56:27.117 slightly more skeptical in the long run economics of of the big players. On the 00:56:27.200 --> 00:56:30.797 other hand, maybe it makes you like more bullish about the startup ecosystem. Uh, 00:56:30.880 --> 00:56:33.597 it certainly should make you more bullish about uh startup application 00:56:33.680 --> 00:56:36.317 companies, right? being able to do interesting things, which is why we're 00:56:36.400 --> 00:56:39.917 so excited about that. Um, you know, it should make you probably, you know, a 00:56:40.000 --> 00:56:44.477 bit more excited about about certainly about China. Um, 00:56:44.560 --> 00:56:47.357 on the other hand, the Chinese competition putting pressure on the 00:56:47.440 --> 00:56:49.997 American system to not screw itself up is very positive. So, it should probably 00:56:50.080 --> 00:56:53.597 make you a little bit more bullish on the US. Um, and so, yeah, I think, you 00:56:53.680 --> 00:56:56.797 know, the these are, yeah, these are yeah, these are are live dynamics and I 00:56:56.880 --> 00:56:59.757 I think we still need more time to pass before we know the exact answer. I 00:56:59.840 --> 00:57:01.997 should say this, but sometime because sometimes I don't sometimes I freak 00:57:02.080 --> 00:57:06.237 people out when I say these are open questions. Um, when a company is 00:57:06.320 --> 00:57:09.757 confronted with fundamentally open strategic or economic questions, it's 00:57:09.840 --> 00:57:14.157 often a big problem because a company needs to have a strategy and the 00:57:14.240 --> 00:57:18.077 strategy needs to be very specific. Um, and a company has to make like very 00:57:18.160 --> 00:57:21.837 specific concrete choices about where it like deploys investment dollars and 00:57:21.920 --> 00:57:25.277 personnel and like the strategy has to be like logical and coherent or the 00:57:25.360 --> 00:57:26.637 company kind of collapses into chaos. 00:57:26.720 --> 00:57:29.597 And so like companies like need to answer these questions and if they get 00:57:29.680 --> 00:57:30.797 the answers wrong, they're really in 00:57:30.880 --> 00:57:34.637 trouble. Um, venture We have our issues 00:57:34.720 --> 00:57:38.317 and venture but a huge advantage that we have is we don't have to we we can bet 00:57:38.400 --> 00:57:42.397 on multiple strategies at the same time right um and and we are doing this so we 00:57:42.480 --> 00:57:45.437 are betting on big models and small models and prepared train models and 00:57:45.520 --> 00:57:48.797 open source models right and and you know and foundation models and 00:57:48.880 --> 00:57:53.677 applications right uh and consumer and enterprise and so the portfolio approach 00:57:53.760 --> 00:57:57.997 the nature of it is like we we are aggressively basically uh we we are 00:57:58.080 --> 00:58:01.277 aggressively investing behind every strategy that we've identified that we 00:58:01.360 --> 00:58:04.637 think has a plausible chance of even when that even when that's 00:58:04.720 --> 00:58:07.917 contradictory to another strategy that we're investing in and one is just like 00:58:08.000 --> 00:58:10.637 the world's messy and probably a bunch of things are going to work and so like 00:58:10.720 --> 00:58:13.517 there's not going to be clean yes or no answers to a bunch of this like a lot a 00:58:13.600 --> 00:58:15.757 lot of the answers to this I think are just going to be and answers but the 00:58:15.840 --> 00:58:19.117 other is like if one of these strategies doesn't work like you know we're not 00:58:19.200 --> 00:58:22.637 we're not trying to hedge per se but you know we're going to have representation 00:58:22.720 --> 00:58:25.357 in the portfolio of the alternate strategy and and so we're going to have 00:58:25.440 --> 00:58:28.717 mult multiple ways to win. So anyway, that's that's the goal. That's the 00:58:28.800 --> 00:58:32.317 theory of why we are, you know, kind of taking the approach in the space that 00:58:32.400 --> 00:58:35.757 we're taking. Um, and that's why I have a big smile on my face when I say that 00:58:35.840 --> 00:58:37.757 there are these big open questions because I think that actually works to 00:58:37.840 --> 00:58:42.237 our advantage. >> It's a good seg uh to A16Z questions 00:58:42.320 --> 00:58:45.997 because we we've gotten a few in so far and and uh we had a few that uh were 00:58:46.080 --> 00:58:50.877 were sent in ahead as well. So uh I'll start one with a with a broad topic. 00:58:50.960 --> 00:58:52.717 What is something you and Ben disagree 00:58:52.800 --> 00:58:57.997 and commit on? disagree commit. Um, you know, we agree. 00:58:58.080 --> 00:59:00.557 I mean, we we Ben I was going to say, you know, we're an old married couple, 00:59:00.640 --> 00:59:05.197 so we argue argue constantly, but we've been >> where the romance is dead. 00:59:05.280 --> 00:59:07.357 >> The romance is long dead. Yes. Yes. Yes. 00:59:07.440 --> 00:59:08.957 Yes. The light the fire the fire has 00:59:09.040 --> 00:59:14.157 long since gone out. Um, but um uh yes, 00:59:14.240 --> 00:59:15.837 if you Yes. We're in the park squabbling 00:59:15.920 --> 00:59:20.557 all the time. Um so, um yeah, I mean, so 00:59:20.640 --> 00:59:23.197 look, we debate everything. We we argue about everything. We that that said like 00:59:23.280 --> 00:59:25.517 you know one of the things that's made our partnership work is like we do we do 00:59:25.600 --> 00:59:28.637 tend to come to the same conclusion like each of us is open to being persuaded by 00:59:28.720 --> 00:59:31.197 the other one and so we we end up coming you know we end up coming to the same 00:59:31.280 --> 00:59:34.637 conclusion most of the time. Um so I would say there there aren't like a 00:59:34.720 --> 00:59:37.757 there aren't I said specifically sitting here today there are like zero issues 00:59:37.840 --> 00:59:40.637 where I'm sitting here and I'm like I can't believe you know I just I can't 00:59:40.720 --> 00:59:43.917 believe I'm you know I'm putting up with this crazy thing on on his on his part 00:59:44.000 --> 00:59:46.637 that he's doing um that I really disagree with but I feel like I have to 00:59:46.720 --> 00:59:48.797 commit to or I I don't think vice versa. 00:59:48.880 --> 00:59:50.717 Um and so so we don't have any of those. 00:59:50.800 --> 00:59:54.477 Um, you know, quite honestly, the biggest thing I say the biggest thing 00:59:54.560 --> 00:59:59.197 that I that he and I the biggest thing that he and I discuss, this this by the 00:59:59.280 --> 01:00:01.597 way, this is not this is not the most important thing we're doing, but it is a 01:00:01.680 --> 01:00:02.957 topic since somebody asked the question. 01:00:03.040 --> 01:00:06.317 The biggest thing he and I discuss where I I don't know, maybe I'm always like 01:00:06.400 --> 01:00:09.197 second guessing myself or I I I never quite know where I should come out on it 01:00:09.280 --> 01:00:13.037 that he and I talk about a lot is just like basically the public footprint of 01:00:13.120 --> 01:00:17.357 the company. Um so like our pres our presence in the our presence in the 01:00:17.440 --> 01:00:19.757 world in terms of like public statements 01:00:19.840 --> 01:00:23.437 uh controversy um uh you know uh how we 01:00:23.520 --> 01:00:27.597 vocalize and express our views on things um and I would just say there like you 01:00:27.680 --> 01:00:30.237 know there's a real there's a tension there's a real it's you know maybe 01:00:30.320 --> 01:00:33.917 obvious but like a very important tension like generally speaking the more 01:00:34.000 --> 01:00:37.277 out there we are and the more outspoken we are and the more controversial we are 01:00:37.360 --> 01:00:39.677 the better for the better for the business in the sense of the 01:00:39.760 --> 01:00:43.197 entrepreneurs love it. Uh the the the 01:00:43.280 --> 01:00:45.677 founders want to work with is very clear 01:00:45.760 --> 01:00:50.317 at this point. The founders want to work with uh uh people who basically are 01:00:50.400 --> 01:00:54.077 brave and controversial and take controversial stands uh and articulate 01:00:54.160 --> 01:00:56.397 things clearly and and they want that for a bunch of reasons. One is because 01:00:56.480 --> 01:00:59.277 it's a demonstration of courage which they appreciate. But the other is 01:00:59.360 --> 01:01:03.677 because it it it it teaches them who we are before they even meet us. Um and and 01:01:03.760 --> 01:01:07.037 and that has just proven to be just like this incredible competitive advantage. 01:01:07.120 --> 01:01:10.077 you know, long long-term LPs will know like this is why we started with a very 01:01:10.160 --> 01:01:13.357 active marketing strategy from the very beginning and like it completely worked. 01:01:13.440 --> 01:01:16.637 Like the the whole thing was if we're able to broadcast our message and we're 01:01:16.720 --> 01:01:20.077 able to basically be very clear in what we believe even to the point where it's 01:01:20.160 --> 01:01:22.957 controversial, like the best founders in the world are going to understand us 01:01:23.040 --> 01:01:25.597 before they even walk in the door, right? And they're going to they're 01:01:25.680 --> 01:01:28.317 going to know us even before they've met us as opposed to everybody else in 01:01:28.400 --> 01:01:31.117 venture, at least at the time, that was basically just like keeping everything 01:01:31.200 --> 01:01:34.637 quiet. Um where they, you know, the founder just has no idea who these 01:01:34.720 --> 01:01:37.117 people are and what they believe. And so that that like worked incredibly well. 01:01:37.200 --> 01:01:41.197 It continues to work incredibly well. Um it's by the way it's you know it's 01:01:41.280 --> 01:01:43.917 generally true across the industry. It's it's it's like generally the case. On 01:01:44.000 --> 01:01:47.437 the other hand, there are externalities to being you know publicly visible and 01:01:47.520 --> 01:01:49.917 and and and to being controversial um on 01:01:50.000 --> 01:01:54.717 many fronts. Um we are I would say this we are we're very much we're trying very 01:01:54.800 --> 01:01:57.277 hard to thread this needle. So like we're we're not backing off of generally 01:01:57.360 --> 01:02:00.237 being a a company that does a lot of outbound. we, you know, we Eric 01:02:00.320 --> 01:02:02.797 Worenberg and the team that he's built, you know, that we've talked to you guys 01:02:02.880 --> 01:02:06.637 about in the past, um, you know, is I is already off to the races. Um, you know, 01:02:06.720 --> 01:02:09.037 we're we're going to, you know, we're tripling down on the idea of basically 01:02:09.120 --> 01:02:12.317 being the leaders and articulating the tech and business issues that matter. 01:02:12.400 --> 01:02:14.957 You know, the, you know, the issues for sure that people need to be able to 01:02:15.040 --> 01:02:19.117 understand. Um, and and that's proven to be very effective. By the way, a fair 01:02:19.200 --> 01:02:23.197 amount of our coms are actually aimed at Washington. Um because again it's like 01:02:23.280 --> 01:02:27.517 if you're a policy maker in Washington and you're sitting there 3,000 mi away 01:02:27.600 --> 01:02:30.477 and your entire information source is like East Coast newspapers that hate 01:02:30.560 --> 01:02:34.557 Silicon Valley. Like that's bad. Um and so you know our ability to like 01:02:34.640 --> 01:02:38.317 broadcast, you know, inform points of view on technology. We just we meet 01:02:38.400 --> 01:02:41.597 people in DC all the time um who say, "Yeah, I you know, most of what I know 01:02:41.680 --> 01:02:43.917 about this topic I learned from you guys because I listened to the podcast, I 01:02:44.000 --> 01:02:46.797 read the articles, I watched the YouTube channel." Um and so, you know, we're 01:02:46.880 --> 01:02:49.837 we're going to continue to do that. And so we, you know, over over over overall 01:02:49.920 --> 01:02:52.557 we have a, you know, we're kind of on our front foot on that stuff. But yeah, 01:02:52.640 --> 01:02:55.997 he he and I do he and I do go back and forth a bit on exactly how, yeah, how 01:02:56.080 --> 01:02:57.597 many third rail topics should we touch? 01:02:57.680 --> 01:03:01.597 Um, and uh and how frequently. And I I would say we're we're we are trying to 01:03:01.680 --> 01:03:03.117 we are trying to moderate that. 01:03:03.200 --> 01:03:07.197 >> As Elizabeth Taylor said, as long as I spell our name right, um, it's 01:03:07.280 --> 01:03:09.997 oftentimes could be good in most 01:03:10.080 --> 01:03:14.797 scenarios, particularly when it comes to little tech. uh double uh and also I 01:03:14.880 --> 01:03:18.877 think embedded in that question is probably uh some degree of of uh uh the 01:03:18.960 --> 01:03:22.317 relationship that you and Ben have which is now going on 30 plus years at this 01:03:22.400 --> 01:03:24.317 point. Uh so much so that that Mark has 01:03:24.400 --> 01:03:27.757 become uh one person representing both 01:03:27.840 --> 01:03:29.837 uh some people refer to Mark as Andre 01:03:29.920 --> 01:03:35.197 and Horowitz no lost the mark have combined just into one person. Uh 01:03:35.280 --> 01:03:37.677 >> yes >> that's the result of 30 plus years 01:03:37.760 --> 01:03:41.437 working together. Okay. Um, so it's been 2 years since you've reorganized around 01:03:41.520 --> 01:03:45.917 AI, launched AD. What do you think you got most right? Uh, and in hindsight, is 01:03:46.000 --> 01:03:51.037 there anything that you underestimated or or missed in that decisioning process? 01:03:51.120 --> 01:03:52.717 >> No, I mean, look, we made we made plenty 01:03:52.800 --> 01:03:55.597 of mistakes. I think those were I think those were the right calls. I mean, AI 01:03:55.680 --> 01:03:59.917 was like I said, like you know, the whole theor back up the whole theory of 01:04:00.000 --> 01:04:02.637 venture the whole theory of venture that we've had from the beginning is that you 01:04:02.720 --> 01:04:06.797 know, many people before us have had as well. that's very correct I think is the 01:04:06.880 --> 01:04:09.517 whole theory is like the money adventure is made when there's like a a 01:04:09.600 --> 01:04:12.557 fundamental architecture shift like when there's like a fundamental change in the 01:04:12.640 --> 01:04:16.077 technology landscape. Um and and that's been true for you know adventure 01:04:16.160 --> 01:04:19.677 basically forever. Um uh and and the reason is because if you have a 01:04:19.760 --> 01:04:23.037 fundamental change in technology then you have this period of creativity in 01:04:23.120 --> 01:04:26.237 which you can have basically aggressive you know very aggressive kind of people 01:04:26.320 --> 01:04:28.717 you know kind of start these new companies and and they have this kind of 01:04:28.800 --> 01:04:32.077 shot to kind of come in and you kind of win categories before big companies can 01:04:32.160 --> 01:04:35.997 respond. um if there's no fundamental change in technology, it's very hard to 01:04:36.080 --> 01:04:38.557 make startups work because the big companies just end up doing everything. 01:04:38.640 --> 01:04:42.637 And so you so venture kind of, you know, sort of lives or dies on on the basis of 01:04:42.720 --> 01:04:47.037 these of these waves of these transitions. Um and and so there's 01:04:47.120 --> 01:04:49.757 always there there's always this question. It's always this question. I 01:04:49.840 --> 01:04:53.437 mean, I would just say the best venture capital firms in history, I I think are 01:04:53.520 --> 01:04:56.637 the ones that were the most aggressive at being able to navigate from wave to 01:04:56.720 --> 01:04:59.517 wave, right? And and and look, I was a beneficiary of this when I came to 01:04:59.600 --> 01:05:03.677 Silicon Valley in ' 904. you know that there was no venture firm in 1994 that 01:05:03.760 --> 01:05:07.277 was like the internet venture capital firm like that it just didn't exist. Um, 01:05:07.360 --> 01:05:10.317 but there were a set of venture capital firms at the time, you know, at the time 01:05:10.400 --> 01:05:13.757 our our firm Kleiner Perkins that said, "Oh, this is a new architecture. This is 01:05:13.840 --> 01:05:17.197 a new technology change. It seems totally crazy. Everybody says you can't 01:05:17.280 --> 01:05:18.717 make money on it. Whatever, whatever. 01:05:18.800 --> 01:05:22.317 These kids are nuts." But like, we're going to make those bets. Um, and so 01:05:22.400 --> 01:05:25.037 they were willing to invest. And by the way, you know, KP in the in the in the 01:05:25.120 --> 01:05:28.797 '90s invested not only in us, but also in Amazon and then Google and like in, 01:05:28.880 --> 01:05:31.757 you know, company after company after company. They invested in at home, which 01:05:31.840 --> 01:05:33.917 basically made made home broadband work. 01:05:34.000 --> 01:05:37.037 um you know they invested in in a fleet of companies and they were a venture 01:05:37.120 --> 01:05:40.797 capital firm that had started in the 1970s around really around what was at 01:05:40.880 --> 01:05:43.757 the time called Minicomputers which was like a you know three generations of 01:05:43.840 --> 01:05:47.757 techn technology back and they had navigated from wave to wave um and and 01:05:47.840 --> 01:05:49.997 you know the same thing is true for Sequoia the same thing is true for 01:05:50.080 --> 01:05:53.197 basically any successful venture firm has been in business for you know 30 or 01:05:53.280 --> 01:05:57.437 40 or 50 years and so I I think in this business like of all businesses like you 01:05:57.520 --> 01:06:01.837 you just you need you need to get onto the new thing um you know it it was I 01:06:01.920 --> 01:06:05.837 mean quite honestly it was I pretty amazing that most of the venture 01:06:05.920 --> 01:06:07.597 ecosystem just decided to sit crypto 01:06:07.680 --> 01:06:11.357 out. Um and and the number of VCs that 01:06:11.440 --> 01:06:14.717 we talked to between call it, you know, the release of the Bitcoin white paper 01:06:14.800 --> 01:06:18.557 in 2009 to the beginning of the crypto war in 2021 who just basically said, 01:06:18.640 --> 01:06:22.557 "Oh, we're not going to do crypto." It was fairly it's I I like I don't I I 01:06:22.640 --> 01:06:24.797 never quite know what to do with the VC who says, "Oh, there's a new wave of 01:06:24.880 --> 01:06:27.197 technology and I'm very deliberately not going to participate in it." And I'm 01:06:27.280 --> 01:06:30.957 always like like, "Is that not the job?" 01:06:31.040 --> 01:06:34.957 Right? Like so so so like I was fairly amazed by the VCs that didn't make the 01:06:35.040 --> 01:06:38.717 jump uh to crypto. You know they they looked briefly smart during the crypto 01:06:38.800 --> 01:06:41.517 wars I would say of the last you know three or four years and I think they 01:06:41.600 --> 01:06:45.357 they probably look maybe a little bit less smart now. Um you know AI is 01:06:45.440 --> 01:06:47.837 another one of these where there are certain firms that are are jumping all 01:06:47.920 --> 01:06:49.917 over it and there are certain firms that are just kind of sitting back and 01:06:50.000 --> 01:06:53.917 letting it happen. Um and um and and by the way there were certain firms that 01:06:54.000 --> 01:06:55.997 never made it to the internet. I mean there were there were firms that were 01:06:56.080 --> 01:06:59.197 very well known in the 80s um and very successful that just like did not make 01:06:59.280 --> 01:07:01.917 the jump uh to the internet and basically just petered out. And so 01:07:02.000 --> 01:07:04.637 anyway long-winded way of saying I think I think in this business of all 01:07:04.720 --> 01:07:07.757 businesses you have to jump you have to jump on the new wave. Um and I and I 01:07:07.840 --> 01:07:10.397 think we got the magnitude of it of it right that this is like a fundamental 01:07:10.480 --> 01:07:14.317 fundamental transformation inside the firm. Um you know AD is you know AD is 01:07:14.400 --> 01:07:16.877 doing great. Um AD AD itself I believe 01:07:16.960 --> 01:07:21.757 is also a beneficiary of AI. um right because in in two ways one is a lot of 01:07:21.840 --> 01:07:25.357 the kinds of products that AD companies build themselves benefit from AI and 01:07:25.440 --> 01:07:27.757 then also AI is a driver of demand in 01:07:27.840 --> 01:07:32.557 other sectors of AD like like energy and materials. Um and so I you know I think 01:07:32.640 --> 01:07:35.997 that that generally is is very consistent and you know is working well. 01:07:36.080 --> 01:07:41.037 Um by the way you know crypto's back back to being a you know I would say an 01:07:41.120 --> 01:07:44.637 exciting industry as a consequence of all the policy changes. Um and then and 01:07:44.720 --> 01:07:46.957 then there's even going to be I think intersections. I I think there's 01:07:47.040 --> 01:07:50.157 actually going to be quite a few intersections between AI and crypto. Um 01:07:50.240 --> 01:07:54.157 and then and then biote you know biotech also bio and healthcare I think are 01:07:54.240 --> 01:07:57.197 obviously going to be transformed by AI both on the healthcare side and on the 01:07:57.280 --> 01:08:00.797 actual drug discovery side and you know and that's underway. And so any anyway 01:08:00.880 --> 01:08:04.157 so like the the the individual efforts in the firm feel good um and suitable 01:08:04.240 --> 01:08:08.717 for the time the inter the interactions between the teams um and the kind the 01:08:08.800 --> 01:08:11.917 the hybrid ideas you know the companies that are coming at these things from 01:08:12.000 --> 01:08:16.317 multiple angles uh you know feels really good um you know maybe the correlarying 01:08:16.400 --> 01:08:19.037 question is like you know what do we feel like we're missing right now um and 01:08:19.120 --> 01:08:22.637 I I think the answer is really not like I don't I don't think like right now 01:08:22.720 --> 01:08:26.077 we're not missing a vertical like I I don't like as of right now like there 01:08:26.160 --> 01:08:29.117 there's not like a specific vertical of like I don't know whatever that like 01:08:29.200 --> 01:08:31.437 where we just like, oh, we just need, you know, we need the equivalent of a 01:08:31.520 --> 01:08:34.397 new of a new unit or the equivalent of a new um you know, new fund or whatever. I 01:08:34.480 --> 01:08:37.597 don't I don't see that at the moment. I think it's more executing extremely well 01:08:37.680 --> 01:08:40.477 in the verticals that we have in front of us. Um and um and then, you know, 01:08:40.560 --> 01:08:43.917 being the best possible partner to the to the portfolio companies. 01:08:44.000 --> 01:08:46.157 >> Yeah. Actually, on the point of of AD, 01:08:46.240 --> 01:08:49.997 um because uh AI is creating and there's 01:08:50.080 --> 01:08:52.317 a lot of talk around AI taking jobs, 01:08:52.400 --> 01:08:54.957 etc. Ironically enough, the jobs in AD 01:08:55.040 --> 01:08:59.757 sectors have never been more in demand in the physical world related to energy, 01:08:59.840 --> 01:09:03.277 related obviously to data center build, etc. So like the the pendulum it seems 01:09:03.360 --> 01:09:05.757 like also is uh is swinging from just an 01:09:05.840 --> 01:09:10.397 accelerant standpoint from from a society uh point of view. Um you talked 01:09:10.480 --> 01:09:13.997 about the importance of society also needing to be ready for tech adoption. 01:09:14.080 --> 01:09:17.517 Like have you seen that accelerating of recently? what's your sentiment of of 01:09:17.600 --> 01:09:20.877 how to actually um increase that just to 01:09:20.960 --> 01:09:25.037 also make sure the convergence of of adoption also falls in line with with 01:09:25.120 --> 01:09:27.517 how quickly tech is is actually being implemented. 01:09:27.600 --> 01:09:31.037 >> Yeah. So, you know, look, we've talked about this before, but um you know, 01:09:31.120 --> 01:09:32.637 look, for a very long time, tech was 01:09:32.720 --> 01:09:36.157 just not a very relevant look, if you go 01:09:36.240 --> 01:09:39.997 back over like whatever 300 years, like there's just like recurring waves of 01:09:40.080 --> 01:09:43.757 like total panic and freakout caused by new technology. Or even you go back 500 01:09:43.840 --> 01:09:46.157 years, you go back to the printing press, you know, which basically was 01:09:46.240 --> 01:09:48.717 handin-hand with the the sort of creation of Protest Pro Protestantism, 01:09:48.800 --> 01:09:53.037 which really changed things. Um, and then um, you know, you you go back to 01:09:53.120 --> 01:09:55.437 um, you know, there there were just always kind of, you know, continuous 01:09:55.520 --> 01:09:57.837 panics there. You know, there have been m there have been multiple ways of 01:09:57.920 --> 01:10:00.477 automation panics for the last 200 years. You know, a lot of the 01:10:00.560 --> 01:10:04.397 foundational panic under Marxism was basically a fear of of of of the 01:10:04.480 --> 01:10:08.637 elimination of jobs through the application of automation. um uh you 01:10:08.720 --> 01:10:10.877 know a lot of the same arguments you hear today about like AI is going to 01:10:10.960 --> 01:10:13.357 centralize all the wealth in a handful of a few people and everybody else is 01:10:13.440 --> 01:10:16.397 going to be poor and emiserated like that that basically is what Markx used 01:10:16.480 --> 01:10:21.037 to say um which I think was by the way wrong then is wrong now we can talk 01:10:21.120 --> 01:10:24.797 about but um you know and then even like in the 1960s there was this whole panic 01:10:24.880 --> 01:10:29.277 around around AI um uh replacing all the jobs there was this there's this great 01:10:29.360 --> 01:10:32.077 uh it's long long forgotten but it was a big deal at the time during the Johnson 01:10:32.160 --> 01:10:35.837 administration you read these AI pause letters today you know that this one 01:10:35.920 --> 01:10:37.037 that just came out a few weeks ago that 01:10:37.120 --> 01:10:39.997 Prince Harry uh headlined of all people. 01:10:40.080 --> 01:10:43.517 Um and um uh uh you know he talks about 01:10:43.600 --> 01:10:48.157 AI is going to ruin everything and it's like and 1964 there was basically a 01:10:48.240 --> 01:10:52.957 group of like the leading lights in academia science and uh you know um kind 01:10:53.040 --> 01:10:56.237 of public affairs that there was this thing called the triple committee or the 01:10:56.320 --> 01:10:59.117 committee for the triple revolution. If you do a Google search on it's like 01:10:59.200 --> 01:11:02.317 committee for the triple revolution Johnson white house or whatever you'll 01:11:02.400 --> 01:11:06.557 this thing will pop up. Um and you know it was a very similar kind of manifesto 01:11:06.640 --> 01:11:09.117 of like we need to stop the march of technology today or we're going to ruin 01:11:09.200 --> 01:11:12.717 everything. Um and and then you know even in the course of the last 20 years 01:11:12.800 --> 01:11:17.117 there was like a big panic around um actually outsourcing in the 2000s was 01:11:17.200 --> 01:11:20.397 going to take all the jobs and then it was actually robots weirdly enough in 01:11:20.480 --> 01:11:23.677 the 2010s which is amazing because robots didn't even work in the 2010s and 01:11:23.760 --> 01:11:27.357 they kind of you know still don't. Um but uh you know there's a panic around 01:11:27.440 --> 01:11:30.957 that and now there's kind of whatever level of AI panic. Um and so like you 01:11:31.040 --> 01:11:33.997 know I would just say like look that you know the way I would describe it is you 01:11:34.080 --> 01:11:37.837 know we in Silicon Valley have always wanted the work that we do to matter. Um 01:11:37.920 --> 01:11:41.437 you know we spend most of our time quite honestly with people telling us that 01:11:41.520 --> 01:11:44.557 everything that we're doing is stupid and won't work. Um like that's the 01:11:44.640 --> 01:11:48.477 default position. Um you know and then basically that flips at some point into 01:11:48.560 --> 01:11:52.317 panic about how it's going to ruin everything. Um you know it's it's easy 01:11:52.400 --> 01:11:55.837 sitting out here to be cynical about that. Um especially when you kind of see 01:11:55.920 --> 01:11:59.517 the patterns over time. I you know my view is we need to be actually very 01:11:59.600 --> 01:12:02.397 respectful of that and we need to be very aware of that and and basically 01:12:02.480 --> 01:12:05.757 that we you know I use the metaphor with the dog that caught the bus like we 01:12:05.840 --> 01:12:07.917 always wanted to work on things that matter we are working on things that 01:12:08.000 --> 01:12:11.677 matter uh people in the rest of society actually really do care about these 01:12:11.760 --> 01:12:14.957 things um and you know and it's our responsibility to think that all through 01:12:15.040 --> 01:12:18.237 very carefully and to do a good job um you know both not just building the 01:12:18.320 --> 01:12:20.957 technology but also explaining it you know look you know I think we have a 01:12:21.040 --> 01:12:24.157 real obligation to uh you know to to really explain ourselves and engage on 01:12:24.240 --> 01:12:27.597 these issues um in terms of how to measure how going you know it's it's 01:12:27.680 --> 01:12:28.957 sort of the classic social science 01:12:29.040 --> 01:12:31.997 question um uh which is like okay if you 01:12:32.080 --> 01:12:36.397 want to understand basically you know patterns of people there's basically two 01:12:36.480 --> 01:12:41.037 ways to understand what people are doing and thinking um one is to ask them and 01:12:41.120 --> 01:12:44.477 and then the other is to watch them um and like every social every social 01:12:44.560 --> 01:12:47.357 scientist like every sociologist will will will will tell you this which 01:12:47.440 --> 01:12:51.517 basically is you can you can ask people right and and the way you do that right 01:12:51.600 --> 01:12:56.797 is like you know surveys focus groups polls um you know what they think Um but 01:12:56.880 --> 01:12:59.197 then but then you can watch them and you can do what's you know called reveal 01:12:59.280 --> 01:13:01.837 preferences. They're just observe behavior which is you can actually watch 01:13:01.920 --> 01:13:04.557 their behavior and and and what you often see in many areas of human 01:13:04.640 --> 01:13:08.237 activity including politics and many different aspects of society and culture 01:13:08.320 --> 01:13:11.037 over time is the answers that you get when you ask people are very different 01:13:11.120 --> 01:13:15.277 than the answers that you get when you watch them. Um and the reason is because 01:13:15.360 --> 01:13:19.037 like I mean you could have a bunch of theories as to why this is the Marxists 01:13:19.120 --> 01:13:22.157 claim that people have false consciousness. the the the the somewhat 01:13:22.240 --> 01:13:24.957 the explanation I believe is just people have opinions on all kinds of things 01:13:25.040 --> 01:13:27.997 particularly when they're in a context where they get to express themselves um 01:13:28.080 --> 01:13:30.877 and they'll have a tendency to kind of express themselves in very heated ways 01:13:30.960 --> 01:13:34.397 and then if you just watch their behavior they're often a lot calmer um 01:13:34.480 --> 01:13:37.437 and a lot more measured and a lot more rational in in what they do and so the 01:13:37.520 --> 01:13:41.517 AI that's playing out in AI right now which is if you pull if you run a survey 01:13:41.600 --> 01:13:45.597 or a poll of what for example American voters think about AI it's just like 01:13:45.680 --> 01:13:48.397 they're all in a total panic it's like oh my god this is terrible this is awful 01:13:48.480 --> 01:13:51.597 it's going to kill all the jobs it's going to ruin thing. The whole thing, if 01:13:51.680 --> 01:13:53.197 you watch the revealed preferences, 01:13:53.280 --> 01:13:56.637 they're all using AI. So, they're like, 01:13:56.720 --> 01:14:01.517 they're downloading the apps. They're using chat GPT in their job. They're, 01:14:01.600 --> 01:14:04.877 you know, having an argument. You You see this online all the time now. I'm 01:14:04.960 --> 01:14:07.197 having an argument with my boyfriend or girlfriend. I don't understand what's 01:14:07.280 --> 01:14:10.797 happening. I take the text exchange. I cut and paste it into chat GPT and I 01:14:10.880 --> 01:14:13.677 have chat GPT explain to me what my partner is thinking and tell me how I 01:14:13.760 --> 01:14:16.557 should answer so that he's, you know, he or she is not mad at me anymore, right? 01:14:16.640 --> 01:14:19.357 So, or like, you know, I have this thing, you know, I have a skin, you 01:14:19.440 --> 01:14:22.317 know, I have a skin condition and doctors, you know, da da da, and I take 01:14:22.400 --> 01:14:26.157 a photo and I and I'm finally like learning about my own health or I use it 01:14:26.240 --> 01:14:29.117 in my job like I, you know, I had to get this report ready for Monday morning and 01:14:29.200 --> 01:14:33.037 I ran out of time and like it, you know, chat GPT really saved my bacon. Um, and 01:14:33.120 --> 01:14:36.957 so people in their daily lives are I would, you know, just you just look at 01:14:37.040 --> 01:14:40.637 the just look at the data you just like they are not only using this technology, 01:14:40.720 --> 01:14:43.757 they love this technology. Um, and they love it and they're adopting as fast as 01:14:43.840 --> 01:14:47.517 they possibly can. So I I tend to think we're going to the public discussion of 01:14:47.600 --> 01:14:49.917 this is going to ping pong back and forth for a while because there is this 01:14:50.000 --> 01:14:53.037 divergence between what people are saying what people are doing. Um but but 01:14:53.120 --> 01:14:56.237 I do think that the what people are doing part is is is obviously the part 01:14:56.320 --> 01:14:59.357 the part ultimately that wins and and and I think this by the way I think this 01:14:59.440 --> 01:15:02.397 technology is going to be exactly the same as every other one. Um which is the 01:15:02.480 --> 01:15:04.557 thing that's going to happen here is this is just going to proliferate really 01:15:04.640 --> 01:15:07.757 broadly. It's going to freak everybody out and then you know 20 years from now 01:15:07.840 --> 01:15:10.717 everybody's going to be like oh thank god we've got it. Like wouldn't life be 01:15:10.800 --> 01:15:15.117 miserable if we didn't have this? um and or you know 5 years from now or or one 01:15:15.200 --> 01:15:19.037 year from now you know people are going to reach that conclusion. Um so I'm I'm 01:15:19.120 --> 01:15:20.637 very optimistic about where this lands. 01:15:20.720 --> 01:15:22.637 It's just that you know there will be turbulence along the way. 01:15:22.720 --> 01:15:24.317 >> I'm I'm smiling because I also witnessed 01:15:24.400 --> 01:15:27.837 that in the wild. Literally late last week I was on the plane. The guy next to 01:15:27.920 --> 01:15:32.157 me was talking to his chat. I could see him and he was like help me draft an 01:15:32.240 --> 01:15:35.677 escalation letter to United for the delay on this flight. I was like sir you 01:15:35.760 --> 01:15:41.240 are on the flight right now. Like at least wait until it's over. 01:15:42.000 --> 01:15:45.037 It was very good though. I'm sure he had a great email crafted as a as a part of 01:15:45.120 --> 01:15:48.237 that. Uh so, okay, I'm going to switch 01:15:48.320 --> 01:15:52.637 gears to uh a few fun questions that that were sent in uh that uh is intended 01:15:52.720 --> 01:15:56.077 to be a lightning round. So, so uh what what is something you've changed your 01:15:56.160 --> 01:15:59.117 mind on recently? Bonus points if it was someone younger than you. 01:15:59.200 --> 01:16:03.597 >> I mean, it's like every day. Um it's just like it's just a constant, you 01:16:03.680 --> 01:16:07.117 know, it's it's almost all like what's in the realm of the possible. Um, I I'm 01:16:07.200 --> 01:16:09.837 I'm terrible at specific examples, so I don't I don't have one like ready at 01:16:09.920 --> 01:16:13.277 hand, but like like I said, it's just it's it's always Yeah. No, it's it's 01:16:13.360 --> 01:16:15.997 often somebody showing up. It's either something somebody writes or something 01:16:16.080 --> 01:16:19.757 somebody says. Um, and yeah, it's almost Yeah, it's very frequently somebody 01:16:19.840 --> 01:16:23.517 who's very young. Um, and um, yeah, it's just like I would say it's a it's a 01:16:23.600 --> 01:16:28.157 routine experience. >> Good way to stay young. Um, do you plan, 01:16:28.240 --> 01:16:32.477 speaking of young, do you plan to be cryogenically frozen? 01:16:32.560 --> 01:16:34.797 Not with current not with current 01:16:34.880 --> 01:16:37.597 cryogenic technology. Um the uh the the 01:16:37.680 --> 01:16:39.677 the track record of that is not great. 01:16:39.760 --> 01:16:44.397 Um uh and um the stories are somewhat horrifying, but uh you know, we'll see. 01:16:44.480 --> 01:16:47.437 >> We'll see. You got we still got some time. 01:16:47.520 --> 01:16:51.757 >> Um how do you stay grounded when your influence itself may distort reality 01:16:51.840 --> 01:16:56.717 around you? >> Yeah. So I was just say the good news, you know, 01:16:56.800 --> 01:16:59.437 I would say the good news on several front. So one is look the concern is 01:16:59.520 --> 01:17:02.557 real. Um, and it's hard for me to it's hard for me to talk about with sort of 01:17:02.640 --> 01:17:06.237 my Midwestern, you know, kind of, you know, Midwesterners, we we either are 01:17:06.320 --> 01:17:09.837 very humble or we we're really good at faking it, but um, uh, you know, it's 01:17:09.920 --> 01:17:12.397 hard to talk about, but requires some introspection. But yeah, I mean, look, 01:17:12.480 --> 01:17:15.437 the the reality warping effect is definitely real. Um, by the way, there 01:17:15.520 --> 01:17:18.717 is a very big advantage to the reality warping effect, um, which is being able 01:17:18.800 --> 01:17:19.837 to get people to do what you want them 01:17:19.920 --> 01:17:23.277 to do. Um, so that, you know, there is 01:17:23.360 --> 01:17:27.597 there is another side to it. Um but it you know it is a concern in terms of 01:17:27.680 --> 01:17:30.557 like having an actual accurate understanding of what's happening. I 01:17:30.640 --> 01:17:33.597 guess I would say two things. I would say one is um you know I mean one is 01:17:33.680 --> 01:17:36.797 just you know my partners I think are quite you know including Ben are quite 01:17:36.880 --> 01:17:39.837 forthright um in telling me when I'm wrong but you know more generally like 01:17:39.920 --> 01:17:43.997 we're just we are very exposed to reality. Um and so and this and again 01:17:44.080 --> 01:17:46.957 you know you mentioned I don't know it's a way to stay younger, make sure their 01:17:47.040 --> 01:17:49.917 hair never grows back or whatever. It's just like you know we run these 01:17:50.000 --> 01:17:52.717 experiments you know cuz we make these decisions about whether to invest or not 01:17:52.800 --> 01:17:55.597 invest and we work with these companies and all their things and like you know 01:17:55.680 --> 01:17:59.357 reality kicks in quickly. You know the the the delusions don't last very long 01:17:59.440 --> 01:18:02.477 in this business. Um because like you know these these things either work or 01:18:02.560 --> 01:18:06.317 they don't. Um and you know you have these like long elaborate you know 01:18:06.400 --> 01:18:08.797 discussions about you know theories on this and that and the other thing and 01:18:08.880 --> 01:18:12.157 then reality just like completely smacks you square in the face you know like you 01:18:12.240 --> 01:18:16.557 idiot right you know like you know what were you you like you know this is like 01:18:16.640 --> 01:18:18.957 the you know the ultimate frustration of the business which is also very 01:18:19.040 --> 01:18:21.357 motivating which is the number of times that you think that you've applied 01:18:21.440 --> 01:18:24.717 superior analysis and then you've either invested or not invested based on that 01:18:24.800 --> 01:18:27.277 analysis and it turns out it was just you the analysis was just completely 01:18:27.360 --> 01:18:31.037 wrong right um and you know you just like completely overrated your ability 01:18:31.120 --> 01:18:33.757 to epistemically you know kind of analyze these things you just you know 01:18:33.840 --> 01:18:37.757 basically inflicted harm like I always the question is always you know it's 01:18:37.840 --> 01:18:41.277 sort of you know any activity that we do is it value add or is it actually value 01:18:41.360 --> 01:18:45.197 subtract right and and and I think in this business of all businesses is kind 01:18:45.280 --> 01:18:49.117 of like that and and that applies to all of my own contributions as well so so 01:18:49.200 --> 01:18:51.757 there is that and then and then I would say um you know maybe the final thing is 01:18:51.840 --> 01:18:55.437 just like I do have the entire internet ready to tell me that I'm an idiot so 01:18:55.520 --> 01:18:58.317 that also 01:18:58.400 --> 01:19:03.680 that also doesn't doesn't hurt and it and it does on a regular basis 01:19:04.080 --> 01:19:08.317 on on the point of uh your alluding to earlier about uh decisions on investing 01:19:08.400 --> 01:19:11.677 in companies. My favorite line I think it was from the uh the Cheeky Point 01:19:11.760 --> 01:19:15.597 interview that you did uh was you know when you invest in a company it doesn't 01:19:15.680 --> 01:19:20.237 go well at least it goes bankrupt right if it does if it does well and it does 01:19:20.320 --> 01:19:24.317 fantastically well you hear about it every single day 01:19:24.400 --> 01:19:29.277 >> for the rest of your life. Yeah. For the next for the next 30 years. 01:19:29.360 --> 01:19:32.317 validity smacking you in the face saying you fool. 01:19:32.400 --> 01:19:35.597 >> You had it. It's literally It's literally you had it in your office. All 01:19:35.680 --> 01:19:39.357 you had to do is say yes. And by the way, and this is the thing 01:19:39.440 --> 01:19:43.117 like every great VC like if you this is this is the stories that you know the 01:19:43.200 --> 01:19:46.957 VCs tell each other. Every great VC basically has this history of like my 01:19:47.040 --> 01:19:50.077 god I had it was in my office. The thing was in my office and I said no and if I 01:19:50.160 --> 01:19:55.117 had just said yes. Um and so it's yeah it's very hard to um yes the constant 01:19:55.200 --> 01:19:57.997 reminders in the Wall Street Journal and on CNBC every day that you made a giant 01:19:58.080 --> 01:20:02.237 mistake um is yes very good very good for the the old humility factor. 01:20:02.320 --> 01:20:06.717 >> Yeah very humbling helps you stay grounded uh all the time. Uh last 01:20:06.800 --> 01:20:11.517 question do you plan to go to Mars if and when that opportunity presents itself? 01:20:11.600 --> 01:20:16.120 >> Probably not. 01:20:16.480 --> 01:20:20.957 >> My subliminal Zoom background wasn't uh sending the positive vibes. This is what 01:20:21.040 --> 01:20:22.237 it >> Well, I'm not even willing to leave 01:20:22.320 --> 01:20:25.837 California. Um, 01:20:25.920 --> 01:20:28.157 so I'm barely willing to leave my house. 01:20:28.240 --> 01:20:32.397 So, um, uh, yeah, I may maybe by maybe by VR. 01:20:32.480 --> 01:20:34.717 >> Yeah. >> Um, and then we'll see what happens. I 01:20:34.800 --> 01:20:37.837 mean, look, having said that, I think Elon's going to pull it off. Um, and so 01:20:37.920 --> 01:20:40.877 I think, you know, I don't know. I don't know. I don't want to predict. This is 01:20:40.960 --> 01:20:43.117 not a prediction, but I, you know, I would not be surprised if within a 01:20:43.200 --> 01:20:47.677 decade there's routine trips back and forth. Um, so, uh, yeah, we may, uh, 01:20:47.760 --> 01:20:50.477 this this may actually become a a practical question. And and by the way, 01:20:50.560 --> 01:20:53.277 I do know a lot of people who are probably going to go, 01:20:53.360 --> 01:20:54.957 >> myself included. Put me on that. 01:20:55.040 --> 01:20:58.317 >> Oh, fantastic. >> The the flights around the world have 01:20:58.400 --> 01:21:05.160 prepared me for the six-month journey to Mars, so I will be just fine.