WEBVTT 00:00:05.000 --> 00:00:06.957 Wow, it's so much brighter out here. 00:00:07.040 --> 00:00:10.397 >> It is. It's, you know, the future is bright. 00:00:10.480 --> 00:00:12.117 >> There's lens flare in the future. 00:00:12.200 --> 00:00:15.757 >> Yeah. So, welcome everybody to the Oscars of 00:00:15.840 --> 00:00:20.477 optimism. That, of course, we're opening up the summit with the king of 00:00:20.560 --> 00:00:24.290 AUTONOMOUS WEAPON SYSTEMS. >> [laughter] 00:00:26.400 --> 00:00:28.997 >> ALL RIGHT. SO, 00:00:29.080 --> 00:00:30.917 I'M GOING TO start with the fact that 00:00:31.000 --> 00:00:33.957 we're awarding three X prizes this week. 00:00:34.040 --> 00:00:37.757 Palmer, congratulations for winning our 00:00:37.840 --> 00:00:41.187 wildfire X prize. >> Thank you so much. 00:00:41.270 --> 00:00:45.037 >> [applause] >> This one's been a long time in in the works. 00:00:45.120 --> 00:00:50.877 >> Let's get Let's get the the the slide up here. So, Palmer won $2.2 million for 00:00:50.960 --> 00:00:52.117 Anduril. >> As if you needed it. 00:00:52.200 --> 00:00:54.797 >> Um a million of 00:00:54.880 --> 00:00:57.597 a million of that from Lockheed, 00:00:57.680 --> 00:01:00.157 which was kind of interesting. 00:01:00.240 --> 00:01:03.957 >> We do a lot of work with Lockheed. We also compete with them, but we 00:01:04.040 --> 00:01:07.237 in the in the defense space, it's you know, we we we compete with their 00:01:07.320 --> 00:01:10.957 fighter jet division and we partner with them on some of their payloads. And so, 00:01:11.040 --> 00:01:13.557 we do a lot of stuff together. But, you me and you have both been trying to get 00:01:13.640 --> 00:01:16.277 this autonomous firefighting stuff going for a long time. 00:01:16.360 --> 00:01:20.677 >> mean, you got this prize announced back in 2019. It's been a long time. 00:01:20.760 --> 00:01:25.037 Anduril's one of our first two products was an autonomous firefighting tank that 00:01:25.120 --> 00:01:28.117 we started building in 2017, finished in 2019. 00:01:28.200 --> 00:01:32.357 >> It's insane that we let fires become conflagrations before we put them out. 00:01:32.440 --> 00:01:36.357 >> And it well, and it's and it's kind of insane that people are saying, "Oh, the 00:01:36.440 --> 00:01:39.357 you know, the technology's finally made it possible." And I don't know, I'm not 00:01:39.440 --> 00:01:42.437 going to correct them too often in public, but it's I mean, this has been 00:01:42.520 --> 00:01:44.437 possible for over a decade. It's >> Yeah. All right. 00:01:44.520 --> 00:01:47.517 >> It's just that it's become politically politically 00:01:47.600 --> 00:01:49.557 impossible to ignore the problem now. 00:01:49.640 --> 00:01:53.797 >> We'll watch a 1-minute video clip of what your team achieved. And if you walk 00:01:53.880 --> 00:01:58.917 us through it here. >> Sure. So, you this was 00:01:59.000 --> 00:02:02.717 an example of what we were doing at the test site where we were 00:02:02.800 --> 00:02:06.717 detecting fires, classifying fires. What you don't really see in here is that you 00:02:06.800 --> 00:02:08.557 also need [music] to reject false fires. 00:02:08.640 --> 00:02:11.077 That was part of the test as well. So, you have to know, you know, who's the 00:02:11.160 --> 00:02:14.317 guy smoking a cigarette? Who's the guy What's What's What's a campfire? What's 00:02:14.400 --> 00:02:17.557 a road flare? And then what's something that shows the characteristics of real 00:02:17.640 --> 00:02:22.037 fire? That's what you need to flag to people. That's what you need to flag to 00:02:22.120 --> 00:02:24.517 emergency services. Uh and then you need 00:02:24.600 --> 00:02:29.757 to match make the type of fire to the right asset that's in the area. And so, 00:02:29.840 --> 00:02:34.397 you want to basically use the most cost-effective thing that can respond 00:02:34.480 --> 00:02:37.277 and get there in time. And if something's growing faster, you might 00:02:37.360 --> 00:02:40.997 need something different. If it's in woods versus on a brush hill, you might 00:02:41.080 --> 00:02:43.997 need something different. And so, you know, it's it's about detection, 00:02:44.080 --> 00:02:48.477 classification, rejection, matchmaking, and then going and and hitting the 00:02:48.560 --> 00:02:52.877 target. And the last thing I'll say is that with most of what we do, 00:02:52.960 --> 00:02:56.557 uh our work with the military, you'd go through many of these same 00:02:56.640 --> 00:03:00.437 steps, but if you find a target, let's say it 00:03:00.520 --> 00:03:03.197 takes you, you know, an hour to an hour 00:03:03.280 --> 00:03:05.837 to to get to somebody versus 10 minutes, 00:03:05.920 --> 00:03:10.317 it usually doesn't make as much of a difference. When you're talking about 00:03:10.400 --> 00:03:15.157 fire, though, the difference between 10 minutes and an hour response, totally 00:03:15.240 --> 00:03:18.157 different tools you're going to need to respond with. You know, 00:03:18.240 --> 00:03:22.237 a bomb is is is a is a bomb whether or not it takes a day or an hour or a 00:03:22.320 --> 00:03:25.637 minute to get there. Uh but when you when when a fire is is is is 00:03:25.720 --> 00:03:29.117 metastasizing, you need to pick the right thing that can put it out before 00:03:29.200 --> 00:03:30.917 it gets too large. >> Amazing. 00:03:31.000 --> 00:03:33.877 >> You know, uh we're going to be announcing something 00:03:33.960 --> 00:03:35.877 in a month's time 00:03:35.960 --> 00:03:39.037 uh that you're funding a new X Prize, 00:03:39.120 --> 00:03:44.157 and uh it's going to be secret until then, but we're going to tease it now. 00:03:44.240 --> 00:03:49.117 Uh another $10 million prize. Would you share share with us what we're going to 00:03:49.200 --> 00:03:53.277 do? >> Sure, you know, uh the 00:03:53.360 --> 00:03:56.477 you know, we we should we should keep the details secret cuz there's there's 00:03:56.560 --> 00:03:58.637 some there's some good twists, but 00:03:58.720 --> 00:04:00.957 suffice to say that I'm not I'm not 00:04:01.040 --> 00:04:04.157 interested in joining the race to push 00:04:04.240 --> 00:04:08.237 AI forward with my money. I think that the 00:04:08.320 --> 00:04:12.237 the artificial intelligence side of models is is is pretty well covered by 00:04:12.320 --> 00:04:16.357 plenty of people. I think we've not been doing enough understanding biological 00:04:16.440 --> 00:04:18.837 intelligence particularly beyond people 00:04:18.920 --> 00:04:23.477 and so this is going to be a prize around understanding how animals think 00:04:23.560 --> 00:04:27.357 and making so that we can communicate with them 10 times better than the 00:04:27.440 --> 00:04:30.677 current state of the art for the current smartest species and not just trained 00:04:30.760 --> 00:04:34.677 animals, but wild animals in their own in their own language. 00:04:34.760 --> 00:04:37.477 >> Amazing. [applause and cheering] >> And I think that there's actually a lot 00:04:37.560 --> 00:04:40.357 that we're going to be able to learn. I mean, if you look at large language 00:04:40.440 --> 00:04:44.357 models, they're kind of this, you know, they're they're they're based on 00:04:44.440 --> 00:04:45.917 humanity's way of thinking and 00:04:46.000 --> 00:04:48.877 communicating distilled and I think that 00:04:48.960 --> 00:04:53.397 there's plenty to learn from all of the other species that are on this planet 00:04:53.480 --> 00:04:58.037 and I wouldn't be surprised if it has applications for AI particularly AI 00:04:58.120 --> 00:05:01.317 where you need to be more efficient, you know, more power 00:05:01.400 --> 00:05:04.637 efficient, more time efficient, you know, 00:05:04.720 --> 00:05:08.957 a different point of view, two heads are better than one, whichever whatever 00:05:09.040 --> 00:05:11.517 metaphor you want to use. >> Love it. Love it. All right, I'm going 00:05:11.600 --> 00:05:14.077 to kick off the first official question here. 00:05:14.160 --> 00:05:20.280 P-doom, you're in the AI industry, the defense industry, what's your P-doom? 00:05:20.600 --> 00:05:23.197 >> I'm just not worried 00:05:23.280 --> 00:05:25.837 about AI dooming 00:05:25.920 --> 00:05:30.357 the I guess I'm not worried about AI doing the killing. I'm very very worried 00:05:30.440 --> 00:05:35.757 about evil people >> Yes. >> using moderately competent AI to do evil things. 00:05:35.840 --> 00:05:39.117 >> Exactly. >> And I am s- 00:05:39.200 --> 00:05:40.957 kind of frustrated by a lot of people 00:05:41.040 --> 00:05:45.477 who are new to the dooming hobby. 00:05:45.560 --> 00:05:49.757 It's It's It's sort It's sort of like how we had to deal with 00:05:49.840 --> 00:05:53.197 the war in Ukraine happens and all of a sudden all these people who have had the 00:05:53.280 --> 00:05:58.397 dumbest opinions you've ever heard about the military for their whole life 00:05:58.480 --> 00:06:01.997 and people who believed we lived at the end of history that we had moved past 00:06:02.080 --> 00:06:06.877 conflict that we lived in a world where you could just write a nasty letter in 00:06:06.960 --> 00:06:10.317 the UN forgetting that that only works if it's backed up by credible threat of 00:06:10.400 --> 00:06:12.717 violence and then all of a sudden they change their minds. You're seeing this 00:06:12.800 --> 00:06:17.637 with AI. People who have not thought about this problem for more than 00:06:17.720 --> 00:06:21.197 a couple years or maybe even just a couple months 00:06:21.280 --> 00:06:25.197 are saying oh my god this is the biggest threat to humanity. These are people who 00:06:25.280 --> 00:06:29.077 have spent almost no time actually analyzing what the largest threats to 00:06:29.160 --> 00:06:32.917 humanity are. They haven't spoken to any of the experts in this area. They 00:06:33.000 --> 00:06:37.317 certainly weren't buying freeze-dried food and building prepper bunkers, you 00:06:37.400 --> 00:06:39.397 know, like like like like me and many of 00:06:39.480 --> 00:06:41.917 my my my buddies who who have who have 00:06:42.000 --> 00:06:44.677 indulged fantasies about what doom could 00:06:44.760 --> 00:06:49.517 look like and also thought realistically about it. And so yeah, the AI doom is 00:06:49.600 --> 00:06:52.517 way down the list for me. I'm I'm I'm I'm a big optimist. I think it's going 00:06:52.600 --> 00:06:57.317 to massively reduce the cost of resource extraction, processing, transformation. 00:06:57.400 --> 00:07:01.077 You're going to see every part of society become subject to the same 00:07:01.160 --> 00:07:05.557 economies of scale and efficiency that automated industries that automated 00:07:05.640 --> 00:07:07.957 without AI have seen. So think like 00:07:08.040 --> 00:07:12.037 agriculture or or textiles. You can buy 00:07:12.120 --> 00:07:17.317 2 weeks of calories for less than 1 hour 00:07:17.400 --> 00:07:20.557 of federal minimum wage labor. That's crazy. 00:07:20.640 --> 00:07:24.437 >> Let's give it up for Palmer here. You're going to see that for everything. 00:07:24.520 --> 00:07:29.757 >> David [applause] >> So I I got a really um pressing question on my mind. I'm kind 00:07:29.840 --> 00:07:32.917 of a feeds and feeds and ships and AI foundation models guy. 00:07:33.000 --> 00:07:37.837 And so when you're putting systems in the field, weapons in the field, uh 00:07:37.920 --> 00:07:41.397 you can you can put the intelligence inside it and it this has always been a 00:07:41.480 --> 00:07:43.837 big problem for the Air Force, like if we're going to start flying drones 00:07:43.920 --> 00:07:46.557 instead of pilots, who's going to unleash the weapons and how's that 00:07:46.640 --> 00:07:50.117 decision going to get made? And the the downside of transmitting it back to 00:07:50.200 --> 00:07:52.997 headquarters is you have latency. 00:07:53.080 --> 00:07:57.117 Uh the upside is you have a lot more compute. Uh what you can put locally in 00:07:57.200 --> 00:08:00.957 a device, in a robot in the field, in a plane is limited, but it's really really 00:08:01.040 --> 00:08:04.277 fast. So what's your philosophy on like, you know, I think a lot of people would 00:08:04.360 --> 00:08:08.357 say, well, are we going to have self-contained decision-making weapons 00:08:08.440 --> 00:08:11.797 wandering around enemy territories or are we going to have like are we going 00:08:11.880 --> 00:08:13.597 to have you know, control at the headquarters? 00:08:13.680 --> 00:08:15.877 >> Yeah, more or less. The the real concern 00:08:15.960 --> 00:08:19.277 isn't so much the latency or or the 00:08:19.360 --> 00:08:24.077 speed of computation. The problem is that if you rely on 00:08:24.160 --> 00:08:25.877 you know, a central a central location 00:08:25.960 --> 00:08:28.517 to do your compute, you now need to have 00:08:28.600 --> 00:08:33.357 command and control that has a good high-bandwidth communication back and 00:08:33.440 --> 00:08:37.197 forth between you and potentially thousands or tens of thousands of active 00:08:37.280 --> 00:08:41.117 assets. And if your enemy jams your communications or intercepts your 00:08:41.200 --> 00:08:44.637 communications or hacks your communications, you're done. 00:08:44.720 --> 00:08:49.797 Uh or worse, they just skip targeting all the stuff in the field and they go 00:08:49.880 --> 00:08:53.757 for the place where it's all being controlled. You know, we saw we saw this 00:08:53.840 --> 00:08:57.517 I there was a there was a there there was there was this documentary that 00:08:57.600 --> 00:09:01.437 George Lucas was working on and and the Trade Federation, they had they had 00:09:01.520 --> 00:09:05.077 their central AI command and control blown up and that was how the Battle of 00:09:05.160 --> 00:09:09.197 Naboo actually resolved was all all the battle droids lost 00:09:09.280 --> 00:09:09.957 >> Oh yeah, they all fell over. 00:09:10.040 --> 00:09:13.717 >> That's right, they all fell over. And so I think that that's that that's the 00:09:13.800 --> 00:09:17.917 practical reality that's driving the need to have autonomy living not back 00:09:18.000 --> 00:09:20.077 home, but forward. There's other areas 00:09:20.160 --> 00:09:22.917 like you go underwater, you can't talk 00:09:23.000 --> 00:09:27.357 to things. And sometimes you could talk to things, but you don't want to. Having 00:09:27.440 --> 00:09:32.557 radio emissions actually exposes you necessarily. And so what what you 00:09:32.640 --> 00:09:34.757 for better or for worse, you're going to need this. I actually don't think it's 00:09:34.840 --> 00:09:37.157 as much of a shift as people think. 00:09:37.240 --> 00:09:40.317 I there's a lot of people who look at this and say, "Oh my god, Pandora's Box 00:09:40.400 --> 00:09:43.277 is being opened. You're going to have weapons that are making their own kill 00:09:43.360 --> 00:09:47.277 decisions." And they don't recognize the reality, which is that that's been the 00:09:47.360 --> 00:09:51.637 case for decades. You look at the You look at the You look at the Aegis system 00:09:51.720 --> 00:09:55.317 on destroyers. It has a fully autonomous mode. You can flip a button and it will 00:09:55.400 --> 00:09:57.717 shoot down anything that comes within 00:09:57.800 --> 00:10:00.277 whatever range you programmed it with 00:10:00.360 --> 00:10:05.117 without any human confirmation. You look at SeaWiz and SeaRAM for shooting down 00:10:05.200 --> 00:10:09.957 incoming missiles, incoming incoming artillery fire, incoming missiles, or 00:10:10.040 --> 00:10:13.237 incoming speedboats. Same thing. You can put them in fully 00:10:13.320 --> 00:10:17.317 autonomous modes. And I mean, it Going back to into Vietnam, we had missiles 00:10:17.400 --> 00:10:18.877 that you could launch from an aircraft. 00:10:18.960 --> 00:10:22.797 They flew over the horizon, tried to identify the right radio emitters that 00:10:22.880 --> 00:10:26.157 they believed corresponded with a surface-to-air missile launch site, and 00:10:26.240 --> 00:10:29.837 then struck them without any The person had to decide, "I'm going to send it in 00:10:29.920 --> 00:10:32.637 this area. They're going to look for this signal, and I know that there's 00:10:32.720 --> 00:10:33.997 this probability they're going to get it 00:10:34.080 --> 00:10:36.117 right." That decision is going to remain 00:10:36.200 --> 00:10:39.837 with people. But, autonomous weapons are 00:10:39.920 --> 00:10:43.557 widely widely There's a whole other talk we're not going to have today, but I 00:10:43.640 --> 00:10:45.957 will I I do have a rant about this where 00:10:46.040 --> 00:10:50.597 actually autonomous weapons have existed for thousands of years. If you look at 00:10:50.680 --> 00:10:54.997 Defined as kind of weapons that can execute the intent of what you 00:10:55.080 --> 00:10:59.157 programmed to do independent from the operator to the engineer. You look at, 00:10:59.240 --> 00:11:02.877 for example, pitfall traps that allow people to walk over them, but then 00:11:02.960 --> 00:11:06.117 capture the commander on his horse when he walks by and it's heavier. You look 00:11:06.200 --> 00:11:09.237 at certain poison traps that would that 00:11:09.320 --> 00:11:14.037 would kill people who had who had certain characteristics, not kill people 00:11:14.120 --> 00:11:16.917 who had other characteristics, or not kill animals. 00:11:17.000 --> 00:11:21.037 You look many types of landmines and sea mines 00:11:21.120 --> 00:11:25.997 and even even torpedoes. And so, it depends on on your exact definition. 00:11:26.080 --> 00:11:29.197 But, even going into myth where people are creating golems that can you go out 00:11:29.280 --> 00:11:33.677 and fight on their fight on their behalf or or defend you defend their city. 00:11:33.760 --> 00:11:37.357 I I guess my argument is we've had autonomous weapons for a very very long 00:11:37.440 --> 00:11:41.437 time. It's just a new set of technology that's that's making them better. 00:11:41.520 --> 00:11:44.877 >> So, technical question on that. The foundation models that go into those 00:11:44.960 --> 00:11:47.757 those remote systems >> Or the pit traps. 00:11:47.840 --> 00:11:51.597 >> Yeah, well so that that right right now the open source models 00:11:51.680 --> 00:11:54.317 are all Chinese. So, I'm pretty sure you're not going to be putting Chinese 00:11:54.400 --> 00:11:58.317 models into remote weapons. Are you going to build Anduril foundation models 00:11:58.400 --> 00:12:01.157 or how how are you going to get AI in those devices? 00:12:01.240 --> 00:12:03.597 >> I mean, we've been doing it since 2017. 00:12:03.680 --> 00:12:08.397 So, we're way ahead of everybody. We're we're a decade beyond that question. 00:12:08.480 --> 00:12:12.317 Um I think that we also don't the foundation models that people are 00:12:12.400 --> 00:12:16.157 building are not the types of models you want to be doing this. They're non that 00:12:16.240 --> 00:12:17.677 they're largely non-deterministic. 00:12:17.760 --> 00:12:21.277 They're non-auditable. It it's it's a very different category. The thing that 00:12:21.360 --> 00:12:25.157 we are building, unfortunately, has very few consumer applications. 00:12:25.240 --> 00:12:30.237 >> Mhm. >> And vice versa. Um there of course are LLM use cases for 00:12:30.320 --> 00:12:34.077 defense that make a lot of sense, but it's not guiding my missile or or 00:12:34.160 --> 00:12:38.637 classifying radar your radar radar radar feeds. I mean, and we we're also we 00:12:38.720 --> 00:12:43.157 can't build models that ingest text and then spit out text. I need to process 00:12:43.240 --> 00:12:47.517 raw data radar data feeds. I need to process sonar data from passive 00:12:47.600 --> 00:12:51.357 listening stations on submarines. That's it's it's pretty highly specialized. And 00:12:51.440 --> 00:12:53.277 I I will take some credit here. 00:12:53.360 --> 00:12:57.677 Uh all these people who are talking about how AI is going to change everything 00:12:57.760 --> 00:13:01.317 they are mostly uh recent converts. 00:13:01.400 --> 00:13:05.957 Anduril Industries, you might notice the acronym is AI. Um so, and that that was 00:13:06.040 --> 00:13:08.037 not that was not a mistake. >> Okay. 00:13:08.120 --> 00:13:12.757 >> Alex, jump in. >> Yeah, I think maybe dovetailing on our 00:13:12.840 --> 00:13:16.077 discussion of driving bench, we're now in an era when it's possible to take 00:13:16.160 --> 00:13:19.957 frontier models, drop them into a robotic embodiment, and get pretty 00:13:20.040 --> 00:13:23.757 stunning results. So, I'd love to ask you a question maybe about the scaling 00:13:23.840 --> 00:13:26.197 hypothesis as it applies to autonomous 00:13:26.280 --> 00:13:30.797 weapon systems or autonomous systems in general. If you have to extrapolate out, 00:13:30.880 --> 00:13:32.477 is there a Moore's law? Is there an 00:13:32.560 --> 00:13:36.037 experience curve for what as obviously 00:13:36.120 --> 00:13:39.877 the capabilities of underlying frontier models, whether they're from Frontier 00:13:39.960 --> 00:13:42.637 Labs proper or from AI, 00:13:42.720 --> 00:13:47.277 as as those progress, do we see any equivalent of a Moore's 00:13:47.360 --> 00:13:52.157 law for the autonomous systems themselves? And and maybe the the second 00:13:52.240 --> 00:13:54.077 half of the question would be, do you 00:13:54.160 --> 00:13:56.477 think ultimately we achieve some sort of 00:13:56.560 --> 00:13:59.677 convergency? Anduril it makes a variety 00:13:59.760 --> 00:14:01.237 of embodiments in a variety of different 00:14:01.320 --> 00:14:04.477 form factors. Do you see ultimately some 00:14:04.560 --> 00:14:07.637 general purpose category killer robot 00:14:07.720 --> 00:14:12.637 that is equivalently the generalist robotic embodiment just like we now have 00:14:12.720 --> 00:14:15.957 generalist models that are capable of doing substantially any task? 00:14:16.040 --> 00:14:18.157 >> No, I don't think so on the defense 00:14:18.240 --> 00:14:21.437 side. I think that for manufacturing and 00:14:21.520 --> 00:14:25.757 for hospitality and medicine, you might see this because you get economies of 00:14:25.840 --> 00:14:28.997 scale that uh that that that that that are really that are really material for 00:14:29.080 --> 00:14:33.437 those types of businesses and you know, a humanoid robot may not be the optimal 00:14:33.520 --> 00:14:37.317 form for a burger griller and it may not be the optimal form for a firefighter. 00:14:37.400 --> 00:14:40.717 It may not be the optimal form for a for a nurse [laughter] or an assembly 00:14:40.800 --> 00:14:42.197 technician. >> I see you're smiling. 00:14:42.280 --> 00:14:45.117 >> But even if it's not optimal, the fact 00:14:45.200 --> 00:14:50.237 that it is shared with humanity in its form makes it a really really good form 00:14:50.320 --> 00:14:54.517 factor. And now I can make one robot that does all of these things and I I 00:14:54.600 --> 00:14:56.757 would probably rather have a let's say 00:14:56.840 --> 00:15:00.557 robot that's 5% slower or or you know, 00:15:00.640 --> 00:15:02.677 less reliable than the optimal burger 00:15:02.760 --> 00:15:04.277 flipping robot if I can share 00:15:04.360 --> 00:15:07.797 commonality with that with that nurse. 00:15:07.880 --> 00:15:11.237 In defense, that's not really the case 00:15:11.320 --> 00:15:13.717 because you are in this constant race to 00:15:13.800 --> 00:15:18.317 not just barely beat your adversary, but to massively overpower them. To try and 00:15:18.400 --> 00:15:20.077 make it where you know, the goal is is 00:15:20.160 --> 00:15:22.677 to just barely win a conflict on the 00:15:22.760 --> 00:15:25.997 margins. It is to fight unfair, to have 00:15:26.080 --> 00:15:30.197 overwhelming threat of violence underpinning whatever diplomatic 00:15:30.280 --> 00:15:34.197 overtures or negotiations you have. And and and everyone was always pushing the 00:15:34.280 --> 00:15:37.037 limits. So, if if if I if I have a jet 00:15:37.120 --> 00:15:40.877 that has a reliable 5% every time edge 00:15:40.960 --> 00:15:42.557 over the enemy, that's worth paying a 00:15:42.640 --> 00:15:45.437 lot of money for, right? Like I I I and 00:15:45.520 --> 00:15:49.677 and I I I think the the the loss in performance you would see 00:15:49.760 --> 00:15:54.157 through lack of specialization probably makes it non-viable. Like to to to go to 00:15:54.240 --> 00:15:58.357 the to the extreme like a you know, I build robot submarines and those have a 00:15:58.440 --> 00:16:03.197 specialized form that needs to dive to 6,000 m depth and and sit on the bottom 00:16:03.280 --> 00:16:07.037 of the ocean for months at a time. And then I have other stuff that's on the 00:16:07.120 --> 00:16:10.757 surface and and you know, they're robotic systems that you know, roll up a 00:16:10.840 --> 00:16:13.837 hill and and they loiter and then I have other things that are you know, aerial 00:16:13.920 --> 00:16:16.837 vehicles that are doing very specialized tasks and I just can't imagine a 00:16:16.920 --> 00:16:18.597 convergence that would make sense. 00:16:18.680 --> 00:16:20.837 Unless you're telling me that we're going to get some cool like like 00:16:20.920 --> 00:16:24.717 humanoids that have like you know, on-board fusion reactors and they can 00:16:24.800 --> 00:16:27.757 dive to 6,000 ft depth. >> So, Palmer, why aren't why aren't you 00:16:27.840 --> 00:16:31.997 building those? I mean, if anybody, you should >> [laughter] 00:16:32.080 --> 00:16:35.597 >> Look, you could potentially build such a thing, but then very quickly you realize 00:16:35.680 --> 00:16:38.957 that specialization would allow you to build something even better. You say, 00:16:39.040 --> 00:16:42.237 "Okay, I've built something that can dive to the bottom of the ocean and fly 00:16:42.320 --> 00:16:45.917 to the moon." Man, it could get to the moon a lot faster if it didn't have to 00:16:46.000 --> 00:16:47.237 also dive to the bottom of the ocean. 00:16:47.320 --> 00:16:51.357 So, I think I think defense is going to be this one where kind of like a 00:16:51.440 --> 00:16:54.197 you know, you you look at for example the difference between cars where 00:16:54.280 --> 00:16:58.397 everything has shared platforms within a manufacturer you know, manufacturer and 00:16:58.480 --> 00:17:01.597 they also share a lot of the same standards 00:17:01.680 --> 00:17:05.437 and the interfaces. I think that you compare that with like 00:17:05.520 --> 00:17:09.597 a motorsport where it's all completely different and it's continuously pushing 00:17:09.680 --> 00:17:14.517 forward. I I think the defense side is probably always going to be this this 00:17:14.600 --> 00:17:18.117 weird branch of exotic highly specialized robotic systems. 00:17:18.200 --> 00:17:21.077 >> You know, Palmer, I expected if anybody was going to be building humanoid 00:17:21.160 --> 00:17:22.637 robots, it would be you. >> No, no, no. 00:17:22.720 --> 00:17:24.517 >> just you you said you said absolutely not. 00:17:24.600 --> 00:17:27.477 >> I've I've opted out because um I because 00:17:27.560 --> 00:17:30.637 I believe in them. If I didn't believe that humanoid robotics had such a 00:17:30.720 --> 00:17:34.197 powerful future, then I would probably build them. But I want Anduril to build 00:17:34.280 --> 00:17:36.317 things that wouldn't exist otherwise. 00:17:36.400 --> 00:17:39.880 That's what I generally want to do. 00:17:40.200 --> 00:17:43.957 Correct. All the other people are already doing a great job. Like I I I 00:17:44.040 --> 00:17:47.677 want to use their robots. >> And and and do things with them. Like I 00:17:47.760 --> 00:17:51.037 think humanoids are going to be an incredible general artificial 00:17:51.120 --> 00:17:56.357 intelligence interface between AI and remote systems and a lot of legacy 00:17:56.440 --> 00:18:00.237 platforms that were designed for a person to you know pull the lever, push 00:18:00.320 --> 00:18:03.197 the button, read the screen. Like we're going to be able to take all those and 00:18:03.280 --> 00:18:06.917 automate them without replacing them entirely. So, you have a lot of things 00:18:07.000 --> 00:18:09.997 like radar systems that cost millions, tens 00:18:10.080 --> 00:18:12.277 of millions, even hundreds of millions of dollars. 00:18:12.360 --> 00:18:16.037 Uh nobody wants to man them during a hot conflict cuz they're the first thing 00:18:16.120 --> 00:18:19.917 that's going to go. But if I can put humanoid robots in them, I've now been 00:18:20.000 --> 00:18:23.157 able to automate them without that. So, yeah, I'm not going to build humanoids. 00:18:23.240 --> 00:18:26.917 Uh this is also why Anduril is not building like small quadcopter attack 00:18:27.000 --> 00:18:30.317 drones. Too many other companies doing a great job. There's a there's a lot of 00:18:30.400 --> 00:18:33.357 a lot of areas. And and we're not going to do space launch. Like we do space 00:18:33.440 --> 00:18:36.357 products. We have a lot we operate a lot of our own satellites and a lot of our 00:18:36.440 --> 00:18:39.677 own sensors in space. Uh but I don't want to be a launch provider. That's for 00:18:39.760 --> 00:18:42.597 all the other billionaires to >> Yeah, yeah, let let them handle that. 00:18:42.680 --> 00:18:47.157 That's fantastic. Salim, um you know, when you're operating 00:18:47.240 --> 00:18:49.597 Anduril Industries, you're uh 00:18:49.680 --> 00:18:52.757 decentralizing defense, right? The 00:18:52.840 --> 00:18:57.197 Pentagon and ministries of defense globally have been the most centralized 00:18:57.280 --> 00:18:59.757 entity in the history of humanity. 00:18:59.840 --> 00:19:03.557 Uh you're sitting on the edge dealing with that. Uh 00:19:03.640 --> 00:19:07.477 you must have a massive immune system problem where people inside the entity 00:19:07.560 --> 00:19:09.437 are going, "Nah, we've done that before. 00:19:09.520 --> 00:19:13.437 We can do it better." They either try and suck you in or they try and spit you 00:19:13.520 --> 00:19:16.357 out. How are you navigating that interface because you're doing it in a 00:19:16.440 --> 00:19:22.120 way that I've never seen anybody succeed do it. 00:19:22.440 --> 00:19:26.037 >> I think that this is a problem that's very, very hard at the beginning. 00:19:26.120 --> 00:19:28.447 >> Oh, I said the wrong thing. 00:19:28.530 --> 00:19:30.117 >> [laughter] >> Wow. >> No idea what the 00:19:30.200 --> 00:19:32.957 >> Well, until someone gets on the intercom and tells us to 00:19:33.040 --> 00:19:34.597 >> Someone won a million dollars. 00:19:34.680 --> 00:19:40.600 Congratulations. >> Um what were we talking about? What what what what was it? 00:19:44.200 --> 00:19:47.317 Seriously, I knew the intercom was going to come on. 00:19:47.400 --> 00:19:48.717 What were we talking about again? 00:19:48.800 --> 00:19:51.957 >> Centralization of Pentagon versus decentralized government. 00:19:52.040 --> 00:19:55.877 >> I'm going to I'm going to keep talking till somebody tells me that I can't. Um 00:19:55.960 --> 00:19:59.877 So, we've gotten through this a bit. I think early on there was that concern 00:19:59.960 --> 00:20:03.397 that Anduril was not going to be able to get through that. Our Our hardest 00:20:03.480 --> 00:20:06.437 problem raising money >> Skippy, would you please shut this down? 00:20:06.520 --> 00:20:09.477 >> [laughter] >> I've been through enough of these to to 00:20:09.560 --> 00:20:11.437 know how it's going to work. >> There we go. 00:20:11.520 --> 00:20:14.477 >> I imagine. >> One of our early problems was that our 00:20:14.560 --> 00:20:18.357 investors said, "Palmer, you potential investors." They said, "Look, we believe 00:20:18.440 --> 00:20:20.957 that you'll be able to build things that work. We believe you'll be able to build 00:20:21.040 --> 00:20:25.037 products that are great, but we don't believe the government is capable of 00:20:25.120 --> 00:20:27.757 actually purchasing them at scale." Cuz I mean, you talk you talk about 00:20:27.840 --> 00:20:29.597 centralization, you know about the 00:20:29.680 --> 00:20:32.237 issues. For people who don't, uh until 00:20:32.320 --> 00:20:35.877 just a few years ago, you had about 80% 00:20:35.960 --> 00:20:38.797 of MDAPs, major defense acquisition 00:20:38.880 --> 00:20:40.917 programs. So, the basically anything 00:20:41.000 --> 00:20:44.877 that that that matters, uh 80% of MDAPs 00:20:44.960 --> 00:20:48.197 went to five companies, and that's it. 00:20:48.280 --> 00:20:52.637 And I mean, you've seen incredible consolidation, centralization. Anduril 00:20:52.720 --> 00:20:55.757 has been successful because we use we're we're we're less of a contractor and 00:20:55.840 --> 00:20:58.837 more of a product company. We use our own money to build products, then we 00:20:58.920 --> 00:21:02.437 sell them, and we've gotten to the point now where people in government have done 00:21:02.520 --> 00:21:06.917 very, very well betting on Anduril, and then miraculously we finish on schedule, 00:21:07.000 --> 00:21:09.277 we finish on time, and then the next 00:21:09.360 --> 00:21:14.077 program manager says, "I know how I can make my job easy. I just I just give the 00:21:14.160 --> 00:21:15.917 contract to Anduril." >> Got it. 00:21:16.000 --> 00:21:19.917 >> So, it it's got to take guts to enter the defense industry, right? The most 00:21:20.000 --> 00:21:24.117 incestuous, you know, self-licking ice cream cone on the planet. 00:21:24.200 --> 00:21:25.037 >> Yeah. 00:21:25.120 --> 00:21:28.477 >> And and and what insanity, 00:21:28.560 --> 00:21:31.637 you know, gave you the balls to do that? 00:21:31.720 --> 00:21:35.237 >> The insanity was the reason that I got into it. 00:21:35.320 --> 00:21:39.757 Uh the first page of our pitch deck said Anduril will save taxpayers hundreds of 00:21:39.840 --> 00:21:43.517 billions of dollars a year by making tens of billions of dollars a year. The 00:21:43.600 --> 00:21:46.397 idea being we're not just making products that we're selling, but we are 00:21:46.480 --> 00:21:51.997 displacing the systems that cost needlessly 10 times what they should. 00:21:52.080 --> 00:21:58.037 And Yeah, thank you. >> [applause] >> And so, it does seem crazy, but, you 00:21:58.120 --> 00:22:01.317 know, I don't I I I can't put myself on the 00:22:01.400 --> 00:22:06.077 same pedestal as him, but one of my mentors and idols growing up and then 00:22:06.160 --> 00:22:09.317 also early in my career was John Carmack, the founder of id Software. 00:22:09.400 --> 00:22:11.877 >> John is great. >> John is incredible. And for people who 00:22:11.960 --> 00:22:16.237 don't know, he's basically the inventor of the modern modern 3D game engine, 00:22:16.320 --> 00:22:19.997 first-person shooters, also started a rocket company that competed in the 00:22:20.080 --> 00:22:22.477 >> He he was competing in the X Prize, and I I remember 00:22:22.560 --> 00:22:27.157 >> And he was one of the, if not the only one to be doing reusable 00:22:27.240 --> 00:22:29.277 >> Exactly, he was doing vertical takeoff, vertical landing. 00:22:29.360 --> 00:22:31.277 >> else was even trying to do. He did it. 00:22:31.360 --> 00:22:34.957 >> I remember there's this this name on the side of his vehicles, Armadillo 00:22:35.040 --> 00:22:37.517 Aerospace. How do you pronounce that thing? I 00:22:37.600 --> 00:22:37.997 thought you'd only know. That's right. 00:22:38.080 --> 00:22:40.117 He was using So, yeah, he was using uh he was using uh he was building his own 00:22:40.200 --> 00:22:43.157 sensor fusion algorithms. He was processing them on modern, you know, or 00:22:43.240 --> 00:22:45.637 on you know, commercial-style graphics 00:22:45.720 --> 00:22:50.197 card hardware using his game programming 00:22:50.280 --> 00:22:52.397 experience. Anyway, the point is it was 00:22:52.480 --> 00:22:57.477 called Armadillo Aerospace. That's right. And the thing and yeah, they were 00:22:57.560 --> 00:23:01.037 doing vertical takeoff and landing long before SpaceX was 00:23:01.120 --> 00:23:03.557 was was was I mean, doing anything. 00:23:03.640 --> 00:23:06.037 >> Yeah. No, before it existed, for sure. 00:23:06.120 --> 00:23:10.357 >> And but anyway, the the point is he he 00:23:10.440 --> 00:23:12.677 I I he ended up becoming the CTO of 00:23:12.760 --> 00:23:14.837 Oculus VR. I managed to poach him away 00:23:14.920 --> 00:23:18.517 from Id, but after he left, he went to 00:23:18.600 --> 00:23:21.957 work on AI actually even before I did. 00:23:22.040 --> 00:23:23.557 And so he started he started his own 00:23:23.640 --> 00:23:29.477 company to work on on on AGI. And and I said why why are you why are you 00:23:29.560 --> 00:23:32.517 doing this? >> You know, that the odds of you succeeding and to be clear, I still 00:23:32.600 --> 00:23:35.317 believe AI was going somewhere, but he said, "I'm going to start a company 00:23:35.400 --> 00:23:39.077 that's just going all in on the end state. Like I'm not trying to 00:23:39.160 --> 00:23:42.677 incrementally get there. It's just all in on AGI." And the point that he made 00:23:42.760 --> 00:23:47.477 was, "Palmer, even if I've got a 1% chance of succeeding, which maybe is an 00:23:47.560 --> 00:23:49.557 overestimation, but suppose I have a 1% 00:23:49.640 --> 00:23:55.080 chance of succeeding, and suppose that I could pull it off, 00:23:55.480 --> 00:23:57.877 the the risk-adjusted 00:23:57.960 --> 00:24:00.997 outcome kind of demands that I do it. 00:24:01.080 --> 00:24:02.757 And that was how I felt about defense. 00:24:02.840 --> 00:24:05.477 If there was a 1% chance that I could 00:24:05.560 --> 00:24:09.837 help solve this problem, then it was one that would be massively impactful to 00:24:09.920 --> 00:24:13.597 work on versus, let's say, starting another gaming company, getting into 00:24:13.680 --> 00:24:18.317 space launch, you know, building a car company, which 00:24:18.400 --> 00:24:21.357 it it all of the billionaires, they end up doing the same thing. You don't have 00:24:21.440 --> 00:24:23.797 enough people getting into defense or or 00:24:23.880 --> 00:24:25.957 putting on a bat suit. It's just it's 00:24:26.040 --> 00:24:30.957 just cars and rockets and and they just go in circles like that for for for 00:24:31.040 --> 00:24:33.997 >> So, you know, when I interviewed you on the on the Moonshots podcast last time, 00:24:34.080 --> 00:24:38.597 you mentioned that you're part of a billionaire's boys club chat. 00:24:38.680 --> 00:24:39.637 >> Yeah, yeah. >> [clears throat] >> Oh, yeah. 00:24:39.720 --> 00:24:42.997 >> Are they working on anything worthwhile at this point? I mean, I gave I gave 00:24:43.080 --> 00:24:46.077 them for saying like, "God, you got all this capital. Solve the world's 00:24:46.160 --> 00:24:48.557 biggest problems." >> I think the B boys are doing a lot 00:24:48.640 --> 00:24:50.797 better now than we were a few years ago. 00:24:50.880 --> 00:24:53.277 >> Okay. >> Um >> They're they're stepping up. 00:24:53.360 --> 00:24:56.237 >> And and for people you think we're talking about is I I was giving a lot of 00:24:56.320 --> 00:25:00.917 people a hard time, especially some retired B boys. And uh I was telling 00:25:01.000 --> 00:25:01.837 them what are you what are you doing? 00:25:01.920 --> 00:25:05.917 You have cuz money is one of the resources they have. But one of the 00:25:06.000 --> 00:25:09.357 resources people don't talk about, especially for the retired B boys, is 00:25:09.440 --> 00:25:13.397 that they have credibility, which would allow them to gather people around them, 00:25:13.480 --> 00:25:17.757 teams around them, and capital. Like they could and it's just sitting there 00:25:17.840 --> 00:25:20.197 underutilized if they're doing nothing. 00:25:20.280 --> 00:25:23.597 They have the credibility in that people trust them as a leader to be able to 00:25:23.680 --> 00:25:26.397 lead an organization like that. 00:25:26.480 --> 00:25:30.517 They could have the same ideas and the same words as somebody who's equally 00:25:30.600 --> 00:25:34.637 smart or smarter and they'll still do better for that. And so my point to them 00:25:34.720 --> 00:25:38.597 is you have a responsibility not only as someone who's who's got a lot of capital 00:25:38.680 --> 00:25:41.877 yourself, but who's got this underutilized resource of the ability to 00:25:41.960 --> 00:25:46.557 do more. If if you're not doing that, then are 00:25:46.640 --> 00:25:48.637 you really like what are you even doing? 00:25:48.720 --> 00:25:53.157 Anyone can sit on a boat. Few people can start another multi-billion dollar 00:25:53.240 --> 00:25:54.907 company that does something that matters. 00:25:54.990 --> 00:25:57.957 >> [applause] >> And the good news is most of them I 00:25:58.040 --> 00:26:02.717 think I I I think a lot of people have have gotten the itch to to solve some of 00:26:02.800 --> 00:26:04.597 these problems that are out there. 00:26:04.680 --> 00:26:08.640 Nihilism is is is on is on the down. 00:26:10.050 --> 00:26:11.997 >> [laughter] >> Yeah, that's an interesting point. 00:26:12.080 --> 00:26:15.557 Although Like I said, I think most of the B boys 00:26:15.640 --> 00:26:18.757 I think they seem like they keep pretty busy. 00:26:18.840 --> 00:26:22.037 I don't I don't I don't have a problem with B boys who don't do enough. Like to 00:26:22.120 --> 00:26:25.637 some people they mis- mistake what I what I say like oh Palmer wants other 00:26:25.720 --> 00:26:30.317 billionaires to work on defense or energy or or biology. 00:26:30.400 --> 00:26:34.157 You like my point is not that people need to distract themselves by doing more 00:26:34.240 --> 00:26:37.397 things. A lot of people can only do one thing. 00:26:37.480 --> 00:26:40.197 You know, and that was what I did for a long time. 00:26:40.280 --> 00:26:44.157 It's what I hate is the people who are not doing anything. That's that that is 00:26:44.240 --> 00:26:46.317 that is like a repulsive to me. 00:26:46.400 --> 00:26:50.277 >> So maybe message to the audience, you get to join the B boys slack group or 00:26:50.360 --> 00:26:53.957 whatever this is, Telegram, if you're doing something new 00:26:54.040 --> 00:26:57.517 and differentiated. >> The group actually got shut down and uh 00:26:57.600 --> 00:27:01.477 we've been we've been talking about now that we need to do a B boys 2.0. 00:27:01.560 --> 00:27:04.397 Um but I think I think people post >> some B girls as well. 00:27:04.480 --> 00:27:07.477 >> I think people posted the wrong memes a few too many times. 00:27:07.560 --> 00:27:10.637 >> Boom. >> I I look uh there there was there was a 00:27:10.720 --> 00:27:14.357 time where there were no there were no B girls and then there was this Business 00:27:14.440 --> 00:27:18.317 Insider article uh whining about it and uh the the reality is like it's just 00:27:18.400 --> 00:27:20.917 like me and some guys. >> Like you know, it's not like there's 00:27:21.000 --> 00:27:23.327 this isn't like the Illuminati conference. 00:27:23.410 --> 00:27:25.157 >> [laughter] >> You know, it's like they're they're like 00:27:25.240 --> 00:27:28.797 oh, there's a group and you know, but no women have been invited. I'm like I it's 00:27:28.880 --> 00:27:30.117 not that no women have been invited. 00:27:30.200 --> 00:27:33.357 It's just like me and some some of my boys. But >> [laughter] 00:27:33.440 --> 00:27:37.197 >> uh but actually there there was a there was a there was a there was a girl in it 00:27:37.280 --> 00:27:40.437 by the end. So, uh we need Yeah, we need to we need to restart that. We need to 00:27:40.520 --> 00:27:44.397 restart that group. >> So, you're in Facebook commercials now. 00:27:44.480 --> 00:27:49.117 >> Oh, yeah. Yeah, yeah. >> Yeah, so so the new Meta VR is out, uh 00:27:49.200 --> 00:27:54.077 how does it compare against your vision of what you wanted to achieve? Is it Yeah. 00:27:54.160 --> 00:27:57.677 >> Look, it does a lot of the things that uh it's not exactly what I would make 00:27:57.760 --> 00:28:01.437 were I the guy who's waving the magic wand and and and and you got lots of 00:28:01.520 --> 00:28:04.557 magic wands spinning the CAD, but uh 00:28:04.640 --> 00:28:09.597 look, the new Meta VR glasses are a lot closer in a lot of ways. Uh so, I've 00:28:09.680 --> 00:28:14.557 long been a believer that you need to decouple the compute and the battery 00:28:14.640 --> 00:28:18.357 from the head. You should it should not be that and and I'm not even saying the 00:28:18.440 --> 00:28:19.837 head totally, it's more of the front. 00:28:19.920 --> 00:28:23.917 So, uh before Oculus Quest. So, Oculus Quest was the last thing that I 00:28:24.000 --> 00:28:26.397 worked on before they fired me. And uh 00:28:26.480 --> 00:28:31.637 the prototypes at the time were a prototype series code named Santa Cruz 00:28:31.720 --> 00:28:36.077 and uh we were sure actually showing that publicly before they fired me. And 00:28:36.160 --> 00:28:39.957 uh Santa Cruz had the battery and compute on the back of the head and then 00:28:40.040 --> 00:28:42.597 the light, you know, the part on the soft tissue on the front of your head 00:28:42.680 --> 00:28:46.157 was kept as light as possible. And then at some point after they got rid of me, 00:28:46.240 --> 00:28:49.757 they they jammed it all into one big giant lump in the front and they ruined 00:28:49.840 --> 00:28:53.557 the ergonomics. They forgot why you need a top strap. Uh, 00:28:53.640 --> 00:28:56.557 but you know, it it it it bothers >> Serve them right. 00:28:56.640 --> 00:28:59.037 >> It bothers me most Look, it bothers me 00:28:59.120 --> 00:29:00.957 mostly at this point because uh 00:29:01.040 --> 00:29:03.677 the Oculus Rift CV1, which was the first 00:29:03.760 --> 00:29:06.317 modern consumer VR headset, it got the 00:29:06.400 --> 00:29:10.237 ergonomics more or less perfect in 2016 00:29:10.320 --> 00:29:11.597 and you know, we we were showing it off 00:29:11.680 --> 00:29:15.957 in 2015 and literally nobody has done as 00:29:16.040 --> 00:29:17.797 good of as it and it's been over a 00:29:17.880 --> 00:29:20.437 decade. Every single person screws it 00:29:20.520 --> 00:29:23.117 up, even Apple. And I I I just I don't I 00:29:23.200 --> 00:29:25.637 can't comprehend it. How how could it be 00:29:25.720 --> 00:29:28.317 that the ergonomic design led by a 00:29:28.400 --> 00:29:32.197 19-year-old kid with no experience is 00:29:32.280 --> 00:29:34.357 better than anything that anyone has 00:29:34.440 --> 00:29:39.437 come up with since. Anyway, the new Meta VR glasses are the closest thing. 00:29:39.520 --> 00:29:41.797 And they're very lightweight. They've off-boarded the computer. They've 00:29:41.880 --> 00:29:45.677 off-boarded the batteries. They have really high-quality pass-through. 00:29:45.760 --> 00:29:49.357 Um they they It has a feature that I invented for CV1 called Transformo 00:29:49.440 --> 00:29:51.877 Fabric, which they do not call it. 00:29:51.960 --> 00:29:55.317 The basically using uh fabric uh a a a 00:29:55.400 --> 00:30:00.757 a soft flexible fab black black light-absorbing matte fabric to one 00:30:00.840 --> 00:30:04.437 minimize reflections, but two uh basically seal the cavity between and 00:30:04.520 --> 00:30:08.197 around the lenses. That means that you can move them uh and the fabric just 00:30:08.280 --> 00:30:12.397 stretches to meet where the lenses are for different interpupillary distances 00:30:12.480 --> 00:30:15.997 rather than having like sliding mechanical interfaces and plastic and 00:30:16.080 --> 00:30:20.717 they brought Transformo Fabric back for the first time since my my era. So, I'm 00:30:20.800 --> 00:30:24.677 I'm I'm a huge fan and uh you know, it's it's a third of the price of Apple's 00:30:24.760 --> 00:30:27.597 product and uh somewhat higher higher 00:30:27.680 --> 00:30:32.237 pixel density. I I think that's pretty extraordinary. Oh, and and and 1/8 the 00:30:32.320 --> 00:30:36.357 weight. >> Yeah. Amazing. >> Pretty amazing. Pretty good. 00:30:36.440 --> 00:30:37.917 >> Let's close on Let's close on this. 00:30:38.000 --> 00:30:43.037 We've got, you know, 1,500 entrepreneurs, moonshot visionaries, and 00:30:43.120 --> 00:30:46.037 tens of thousands watching on the live stream. 00:30:46.120 --> 00:30:48.797 Uh what's your advice? 00:30:48.880 --> 00:30:53.917 What, you know, what did you learn in high school that made you succeed at the 00:30:54.000 --> 00:30:57.277 speed you did? We were talking about this on the plane right back. 00:30:57.360 --> 00:31:00.677 >> I I was really lucky to uh I guess maybe 00:31:00.760 --> 00:31:04.317 not lucky. I did it. Uh but >> Well, you are you are lucky. 00:31:04.400 --> 00:31:05.997 >> That's right. >> I am [laughter] lucky. 00:31:06.080 --> 00:31:08.957 >> It's in the name. >> I started an online community called 00:31:09.040 --> 00:31:12.437 ModRetro when I was 14 or 15. Uh it was 00:31:12.520 --> 00:31:14.957 an internet forum for people who were 00:31:15.040 --> 00:31:19.477 modifying vintage game consoles and computers, building handheld units, 00:31:19.560 --> 00:31:22.877 improving the performance in a variety of ways. We had a few thousand active 00:31:22.960 --> 00:31:26.197 members. And uh and a whole team of volunteer 00:31:26.280 --> 00:31:30.157 moderators and administrators with with me me at me at the top of it for some 00:31:30.240 --> 00:31:35.557 reason. And uh I learned a lot of valuable lessons about how you can build an 00:31:35.640 --> 00:31:38.917 organization, how you market an organization, how you get people to 00:31:39.000 --> 00:31:44.317 believe in what you're doing. I learned a lot about dealing with the types of 00:31:44.400 --> 00:31:46.317 problems you deal with when you're running a company, you know. 00:31:46.400 --> 00:31:48.717 >> With humans. >> Well, I mean, you you have you have two 00:31:48.800 --> 00:31:51.397 It's a tale as old as time. You've got two people. 00:31:51.480 --> 00:31:54.877 They're both brilliant. They're both very good at what they do. 00:31:54.960 --> 00:31:56.757 And they hate each other. And they 00:31:56.840 --> 00:32:02.037 disagree over something. You need to somehow mediate the conflict or separate 00:32:02.120 --> 00:32:05.957 them in a way that allows you to get the value that that allows both of them to 00:32:06.040 --> 00:32:10.957 contribute value to the greater project, uh but not have them hate you, not have 00:32:11.040 --> 00:32:15.477 them lose faith in the broader thing, but also you not not kill each other. 00:32:15.560 --> 00:32:18.357 Sometimes that means you need to reconcile things between the two. 00:32:18.440 --> 00:32:21.997 Sometimes it means that you need to put one guy over here and one guy over here. 00:32:22.080 --> 00:32:24.437 And sometimes you need to make sure that they don't know that that's what you're 00:32:24.520 --> 00:32:27.837 doing as you do it. That's just [laughter] I'm not saying that's the key 00:32:27.920 --> 00:32:33.397 to running a successful business is the management of the two problem, but it's 00:32:33.480 --> 00:32:36.427 Yeah, and then and once you make it a three-body problem, now I'm out. 00:32:36.510 --> 00:32:39.357 >> [laughter] >> But uh but but it's it's that and a 00:32:39.440 --> 00:32:42.437 hundred things like it that I learned 00:32:42.520 --> 00:32:46.557 running Mod Retro that I was able to take into starting Oculus, to carry 00:32:46.640 --> 00:32:50.477 forward into what I'm doing in my other companies. And uh that that's why I'm a 00:32:50.560 --> 00:32:54.717 I'm a strong believer that the the the kids should should should get a job. And 00:32:54.800 --> 00:32:57.637 I I I also had a job starting when I was 14, so I learned how the real world 00:32:57.720 --> 00:33:02.837 works, too. But I I So, I'm a huge fan of people doing things at a scale that's 00:33:02.920 --> 00:33:05.317 larger than just themselves. In other words, 00:33:05.400 --> 00:33:07.517 don't just do a project, do a project 00:33:07.600 --> 00:33:09.717 with someone else. Do a project with 00:33:09.800 --> 00:33:14.557 other people. Build in the open, learn the the the the good parts of that, the 00:33:14.640 --> 00:33:17.957 bad parts of that. It it it's been extremely useful to me. And 00:33:18.040 --> 00:33:19.877 >> Has building changed in the age of AI? 00:33:19.960 --> 00:33:22.957 What advice do you have? Again, you know, we've been saying on the pod for a 00:33:23.040 --> 00:33:27.797 while now, you know, forget about grad school, you know, even college is a is a 00:33:27.880 --> 00:33:29.477 maybe. You didn't go to college. You 00:33:29.560 --> 00:33:32.237 know, and high schoolers should what? 00:33:32.320 --> 00:33:35.437 Find a a purpose they care about and use 00:33:35.520 --> 00:33:40.560 AI to go and code it. What's your your closing advice here on that? 00:33:40.880 --> 00:33:42.837 >> You know, one of the reasons I have not 00:33:42.920 --> 00:33:48.597 been as upset as some people about uh 00:33:48.680 --> 00:33:52.407 Now, it's funny. There's something on the screen. It's like it's to not talk about 00:33:52.490 --> 00:33:54.357 >> [laughter] >> Just wanted to make sure you saw that. 00:33:54.440 --> 00:33:56.757 >> I I I did. We're not going to talk about Don't think about it. All right, 00:33:56.840 --> 00:33:59.607 company. All right, great. Great. 00:33:59.690 --> 00:34:03.517 >> [laughter] >> The uh There's the There's a lot of people who 00:34:03.600 --> 00:34:05.117 who who love their craft and they love 00:34:05.200 --> 00:34:07.997 what they do. And uh and and they are 00:34:08.080 --> 00:34:11.157 rightfully mourning uh the end of, for 00:34:11.240 --> 00:34:15.397 example, coding things by hand top to bottom, start to finish as anything but 00:34:15.480 --> 00:34:19.717 like a craft endeavor. Um and and I and I understand that. It's it's the same 00:34:19.800 --> 00:34:23.317 way that a lot of artists felt when the camera was invented. They They whined 00:34:23.400 --> 00:34:26.557 and complained about how it ruined art and it made it so that anybody could 00:34:26.640 --> 00:34:29.797 push a button. And and you know, the these things are always painful for the 00:34:29.880 --> 00:34:31.997 people who are masters of their craft. 00:34:32.080 --> 00:34:35.557 Um I already got screwed in that way a long time ago. When you're a founder, 00:34:35.640 --> 00:34:38.237 you don't get to actually work in any of the tech anyway. All the other people 00:34:38.320 --> 00:34:42.037 get to have all the fun. And so, you know, I I like electrical engineering. I 00:34:42.120 --> 00:34:45.037 like optical engineering. I like mechanical engineering. I don't get to 00:34:45.120 --> 00:34:48.557 do any of it cuz the job of a founder is to 00:34:48.640 --> 00:34:51.677 hire people to to displace them. And and 00:34:51.760 --> 00:34:55.917 and to do what they thought they could do or what they like they do like doing 00:34:56.000 --> 00:34:59.637 better. The job, I've said many times to people, if you want to work on 00:34:59.720 --> 00:35:02.997 technology work for somebody else. If you want to 00:35:03.080 --> 00:35:06.957 work on nonstop start a company. 00:35:07.040 --> 00:35:11.477 And so and so what I would say to people is AI doesn't actually change the 00:35:11.560 --> 00:35:15.237 part of the job very much. You're still on the hook legally. You're 00:35:15.320 --> 00:35:18.357 still on the hook for all the interpersonal stuff. And you could you 00:35:18.440 --> 00:35:21.357 you can delegate some of it, but at the end of the day it's your call, it's your 00:35:21.440 --> 00:35:24.677 decision. It's the same thing for fundraising. It's the same thing for 00:35:24.760 --> 00:35:27.637 general product architecture. And when people make the wrong decisions, it's 00:35:27.720 --> 00:35:31.037 your choice as to when you need to overrule them, 00:35:31.120 --> 00:35:33.317 override them, replace them. I don't see 00:35:33.400 --> 00:35:36.397 AI replacing that part of it anytime 00:35:36.480 --> 00:35:40.637 soon. Not because it couldn't, but because the gains everywhere else are so 00:35:40.720 --> 00:35:45.477 much larger that I think we've got a good 10 years of speed running 00:35:45.560 --> 00:35:48.597 all of the other parts of the business before they before they come to that. 00:35:48.680 --> 00:35:52.237 So, look, AI changes a lot, but I I I'm telling you there's a lot of people you 00:35:52.320 --> 00:35:56.837 are in for a rude awakening if you are like a hobbyist coder or a hobbyist 00:35:56.920 --> 00:35:59.117 engineer and you think you're going to start a company and you think you're 00:35:59.200 --> 00:36:01.237 going to be doing all that stuff? No, 00:36:01.320 --> 00:36:04.117 you're going to learn a hard lesson. It 00:36:04.200 --> 00:36:08.437 Being a founder was was always like AI took your job. It's just that other 00:36:08.520 --> 00:36:11.917 people took it. >> Yeah. Ladies and gentlemen, let's give it up 00:36:12.000 --> 00:36:14.960 for Palmer Luckey. 00:36:19.860 --> 00:36:21.880 >> [music]