WWDC25 Reactions, Apple Under Pressure for AI Progress, Meta in Talks to Scale AI Investments | Casey Handmer, Patrick Mcgee, Aaron Slodov, Gaurab Chakrabarti, Sean Liu, Garth Sheldon-Coulson, Lukas Czinger, Christian Keil, Vlad Matsiiako, Alex Heath

27:57The social media giant has before now mostly depended on its in-house research, plus a more open development strategy to make improvements in its AI technology. Of course, we've been hearing news that Meta is revamping the team behind Llama, really trying to double down on that project. They're not getting out of the game, but they might have kind of like aimed for the wrong target with the particular benchmarks that they were going after. And so had struggle wrangling the behemoth, getting it out the door. But, you know, Zuck's not giving up, of course. He's going to keep working, keep building.

28:32Continuing to cook. And so big tech peers have already invested heavily. Microsoft put more than$13 billion into OpenAI, while Amazon and Alphabet put billions into rival Anthropic. Everyone's got a bet there. Part of those companies' investments have been through credits to use their computing power. where Meta doesn't have a cloud business and it's unclear what format Meta's investment will take. And it's unclear if scale really needs cloud credits. They more, I mean, their cost structure is human labor since they oftentimes take tasks and find experts overseas and increasingly experts in America because a lot of the reinforcement learning is now done on extremely hard math problems.

29:14And so you might need to go to PhD programs to recruit these folks. We should offer maybe physical intelligence or another lab the data from Tyler making this iPhone to train on. Definitely. Because I wouldn't want potential training data to go to waste. Yeah, a body suit so that he's really able to track all the mechanics. We need every digit on his hand tracked so that we can be feedback. This might be the first time an iPhone has been assembled by somebody wearing a tailored suit as well. We're breaking all sorts of records here. CEO Mark Zuckerberg has made AI meta's top priority and said in January that the company would spend as much as$65 billion on related projects this year.

30:00Now, a lot of that capex and a lot of that investment is going into things that are not directly foundation model training, of course. They have reality labs and VR efforts, and they also have a ton of recommendation algorithms that are running on large data centers that they're scaling up. But obviously they're investing tons. The company's push would include an effort to make Llama the industry standard worldwide. Metis AI chatbot, which is already available on Facebook, Instagram, and WhatsApp, is used by 1 billion people per month. Wow, that's a lot. Scale, co-founded in 2016 by CEO Alexander Wang, has been growing quickly.

30:33I believe it's a YC company too. The startup generated revenue of 870 million last year and expects sales to more than double to 2 billion in 2025. Let's go. Let's go. Scale plays a key role in making AI data available for companies because AI is only as good as the data that goes into it. Scale uses scads of contract workers to tidy up and tag images, text, and other data that can be used for AI training. Like the original scale AI business model was labeling and tagging images for self-driving car companies. and then at a certain point I think the self-driving car companies stopped needing so much labeled data and everyone was like, oh, scale AI is not that useful.

31:19Like we don't need them anymore. And then the LLM boom came and the RLHF boom came and it was like, oh, wait, we need 10 times more scale AI stuff. And then the same thing is happening in robotics and the same thing is happening in reinforcement learning more broadly and a ton of different paradigms to move the AI ball down the field. Scale and Meta share an interest in defense tech. Last week, Meta announced a new partnership with defense contractor Anduril Industries to develop products for the U.S. military, including an AI-powered helmet with virtual and augmented reality features. Meta has also granted approval for U.S.

31:54government agencies and defense contractors to use its AI models. I love Defense Llama. It's great. The company's already partnered with Scale on a program called Defense Llama, a version of Meta's Llama Large Language Model intended for military use. Scale has increasingly been working with the U.S. government to develop AI for defense purposes. Earlier this year, the startup said it wanted to contract with the DOD to work on AI agent technology. The company called the contract a significant milestone in military advancement. So, I mean, Scale is such a big company. They're doing a lot of different stuff at this point.

32:26and they're not just doing the business process outsourcing the data generation, but they're also doing stuff like they create evaluations. They have the humanities last exam, the benchmark for AI stuff and then they're also fine-tuning Lama and working with the DoD and Alex is in DC barely often. Anyway, fantastic report from Bloomberg. Go subscribe. Give them some love. Give them some love. we have a post from Pliny the Liberator. Getting a little spicy with Apple. He says, I'm not reading a single AI research paper coming out of that giant stale donut in Cupertino until Siri can do a little bit more than create calendar events on the fourth try.

33:11Shots fired. Apple, they released a research paper last week called The Illusion of Thinking, Understanding the Strengths and Limitations of Reasoning Models via the lens of problem complexity. They basically said that reasoning models are fake and just pattern matching machines. Yes. And I got to defend technology here. I'm a bit of a pattern matching machine myself. I try to see patterns. I try to match them. Yeah. Not unlike a large, you know, reasoning model. Yeah, yeah, yeah. But anyways, I think this is sort of interesting timing. I'll wait to, you know, pass too much judgment. until we see what they release today.

33:55But in some breaking news, Apple just announced that they're basically mimicking the Mac OS experience on iPad. So you can now have multiple apps open at once, which is interesting. Very fun. What else is new in WWC world? iPad OS. They launched new CarPlay and they demoed it on a Mustang. You'll have to see that. There we go. Very funny choice of car, but we love a Mustang. There we go. They're connecting with, you know, core America with that one. They know. People want gas cars. Emily Sundberg is on the ground at WWDC, and she says the tech guy in a hoodie stereotype is dead. She's seeing marine saray sneakers.

34:43I'm actually not familiar. Japanese chore coats and a bunch of other items. tech has pulled the world down to their dress code and now we've all met in the middle yeah so yeah emily will be on the show later this week yeah the ai research paper that they put out is kind of interesting there's this there's this constant drumbeat of like moving the goal posts when when it's like okay so you know now these models can reason and produce like really really thorough research reports and they can definitely follow these patterns and that's super valuable but then people will move the goalposts and say well that doesn't count as thinking like yes we yes we passed the touring test but it's not it's not really thinking it's not really human and uh and and i understand that and i think that there's something right about that but also like it it all doesn't it doesn't really matter because at the end of the day all that matters is like economic impact and like how valuable are these things and how useful are they um at the same time revenue as a benchmark i i think that's the only benchmark that matters um and then and then on the flip side there's this i think the only the only counter to that would be that a lot of useful ai features will be baked into ios in the long run and won't have be directly sure sure sure sure sure sure so maybe you need to look at like user time or something yeah and delight there's a whole bunch of ways but like like the fact of the matter is like like does like is ai really thinking does it matter who cares really like do we need do we need to discuss this uh there's also this funny pattern i i believe it was like either nival or wilmanitis or someone posted how like when the steam engine was invented, there were all these papers saying like, oh, the steam engine is just like the human brain.

36:19And then when electricity was invented, it was like, oh, the network of lights in a city is just like the human brain. Now we understand the brain. And so there's this constant march of like, we created technology and then we're like, oh, this is a perfect representation of how humans work. And maybe we keep missing things. But the question is like, we're clearly knocking down different patterns, like even just long-term memory. Like that has been solved by hard drives for a long time. We've been able to have databases and store lots of data and look up facts on the internet. Like that was something that was uniquely human.

36:50We solved it. It's great. It's not human, but it's great and helpful. And now we can do, you know, next token generation and prediction and predict the next word in a sentence, which is very helpful. And clearly there's something in the human's mind that does that activity. But is it everything a human does? No. Based on training data. Yeah, based on training data. And then reasoning is the same thing, like following a playbook, working through a problem, planning. That's clearly part of the human brain, but it's not the entire thing. But it's still really useful. And so it's all good. More breaking news.

37:23Apple now adding your passport to the wallet app that will be accepted at TSA for domestic flights. Okay. Which is, I guess, valuable. Yeah, yeah. They just Sherlocked Real ID. The government. This is a post from Katie Notopoulos from Business Insider. Yeah, it's funny. I got the passport ID, the passport card. So you can have the paper passport, which people know for flying internationally. And then you have the real ID, the driver's license. And the passport card is something that you can order. and it lets you travel between US, Canada, or US, Mexico without bringing the full passport. And I was super excited.

38:11I was like, oh, this is great. If I'm going to Canada, I'll just bring this thing. One day I was going to Canada. Why were you going to Canada? I have no idea. I have no idea. But anyway, I was going to Canada and I was like, let me bring this passport card. This is great. As I show up, it says, only valid for road and rail crossings. You can't fly with it. Road crossings. Because basically it's a product for truck drivers. that are driving back and forth all day long, and they don't wanna bring a separate passport, they just wanna throw an ID in their wallets, in their Ridge wallets. Next to their RAM.

38:44Yeah, next to their RAM cards. And so I was very disappointed with that product. But hopefully this is better. But I imagine that people will get over their skis and be like, oh, I have my passport in my phone, then they'll eventually try and fly internationally and they'll get caught and then have to send an Uber. I actually wound up sending an Uber to my house getting on the phone with the guy and saying like just break into my house and get my passport and bring it to me at the airport and it worked it was that's great high trust environment

39:16somebody somebody somebody bad is going to be like there's going to be a giant newfoundland dog there he's friendly give him a pat and then and then here's the code word here's the code word Tell him, good boy. Make this sound. Yeah. And he'll let you pass. Yeah, I really have like a massive dog. I still do. Anyway, Vanta. Automate compliance, manage risk, improve trust continuously. Vanta's trust management platform takes the manual work out of your security and compliance process and replaces it with continuous automation, whether you're pursuing your first framework or managing a complex program.

39:51Get on there. Do it. Get on there. Do it right now. Data drop from Andreessen Horowitz, Olivia Moore. has the data. Our team at A16C published benchmarks on revenue growth for AI startups from their proprietary data set. The median B2B AI company is going from zero to$2.1 million in ARR in year one, while the median B2C company is going from zero to 4.2 million ARR in year one. Yes, consumer companies are growing revenue faster. It's pretty remarkable. Even the bottom quartile is doing pretty well. Everyone's making money now. It's just so much different than the dot-com boo. Like, they're, you know, obviously.

40:30Wasn't it? Tons of froth and tests. But still, these are serious numbers. You know, if you start a company within 12 months, you have a couple million bucks in revenue. And it's like, the costs are going to be mostly overhead and employees. Like, you're in a pretty good spot. Has there been a period where B2B categorically was growing much faster than consumer? because I feel like with consumer, sure, with B2B, you have these big contracts which can help grow revenue quickly, but there's also so much friction to get onboarded whereas consumers can just say, yes, I will take this product for$100, and you can do that thousands of times a day.

41:09And so I feel like that part isn't necessarily new, but still really impressive benchmarks. Yeah, I mean, increasingly there are so many ways to get to a paid upgrade. There are more. It's not just that Stripe exists and that in-app payments exist, and there's so many ways to make payments, taking payment frictionless on the internet, which was not the case in 1999. But also just psychologically, people, I think, because of ChatGPT being paid, X has a paid tier. People pay for Patreons. People pay for all sorts of stuff. Subscriptions for games, all sorts of different things for subscriptions now.

41:47Well, part of it is that AI... psychologically even existing gen ai experiences even though they're new are so are oftentimes so magical that people are totally happy to pay because the value is you know the studio ghibli app if they were just monetizing that oh yeah and they had it basically people would have paid a hundred dollars a month for that they would have churned immediately i'm pretty sure facetune is doing like a billion dollars in revenue or like face app like one of those like face swapping like like iOS apps, because they just get you in, give you a free one, they know exactly how to monetize, ask you at the right time, you subscribe, yeah, it's a 14-day free trial, and then it rolls over, and you get charged$99 for the entire year.

42:27Like, they're very good at this stuff. It's very Nikita beer-coded. Anyway, we have Casey Hanmer coming in the studio in the TBP and Ultradome in just a minute. I want to read some of his posts first. He says, 1640 days until China takes the moon. China wants the moon so bad they can taste it. They obsess over it all day. They dream about it all night. They fear the consequences of failure. Viscerally, it's affecting their appetite. They work 80, 90, 100 hours a week. The job isn't done until it's done. Casey says, the last page of our passport is a solemn promise that the USA will guarantee freedom as we lead humanity into an infinite and eternal universe in some desperate corners of the earth.

43:08Totalitarian dictatorships still stomp on the potential of a billion hearts, but the future belongs to freedom. So excited to bring him in and talk to him about that and also about solar energy and manufacturing in America and what's going on with Apple. So welcome to the studio. Casey, how are you doing? I'm very well, thanks. Thanks for having me. Great to have you. As always, what triggered this initial tweet about 1640 days until China takes the moon? I imagine that that's based on a specific mission that they have planned. but break that down because they've done stuff before. They've landed on the moon before with autonomous stuff.

43:49Like, why is this different? Why is this date so important? Yeah, China set a target of being on the moon officially by 2030 with people, Chinese taikonauts, obviously. And they have a space program already. They have a space station. They've had several space stations. They have the ability to fly humans into space and to recover them. And actually, the last news we got about a month ago was that they're looking into ways to accelerate that schedule into 2029. So I picked a date in late 2029, and I thought that's 1 ,640 days away. It's about three and a half years or something from now. It's not that long.

44:22And one thing I can tell you for sure is that the Chinese space program managers fear the consequences of failure in a way that U.S. space program managers have not felt or feared for many a decade. What is what are we should be copying their efforts to NASA or SpaceX or both or some mix? What is the Chinese what's the shape of the Chinese space program right now? Well, I mean, I don't not an expert on the Chinese space program by any means, but it's it's obvious from from the way they work that they're dead serious about it. They are in second place behind SpaceX when it comes to launching stuff into space and they haven't got reusability done yet.

45:01So they're they're producing significantly more cores. and then of course when they're done with them they drop them on villages because they run things differently. Now you know I think that on this show people will understand what I mean when I say the light cone wants to be free. The light cone is a place where humanity needs to expand and experience freedom and the reality is that laws of physics don't guarantee that freedom will succeed. It helps us, freedom helps us perhaps but the challenge we face I think is that NASA is the organization whose job it is to make sure that we don't get a 1958 Sputnik-like surprise ever again.

45:34And there are many, many smart people at NASA working hard on this, but their efforts are somewhat circumscribed and challenged by kind of decades of accumulated craft and problems that are kind of causing major problems within the organization and for the US space program as a whole. I think SpaceX has done a really incredible job of upholding US interest in space. If it wasn't for SpaceX, we'd obviously be in a deep, dark hole by now. But it's a lot to rest on the shoulders of just one company. And I think this has been reinforced in the United States just last few days with people realizing that actually the second place US-based company or whatever is an awful long way behind SpaceX.

46:11It's not for a lack of trying. SpaceX is probably the number one place on Earth for training rocket engineers, and nothing forces them to stay there. Many of them leave and start their own companies, and good for them. But they're still a long way behind. Yeah. Why is the moon so important, in your opinion? There's arguments about security and defense and putting weapons on the moon. There's also resources. like what are the top reasons why we should be going to the moon? I kind of just want to go because it seems fun. Yeah. My son seems excited about it. I want to go with him one day. We need to figure out if it's made of cheese or not.

46:43We do. We do. Yeah. Well, I mean, I don't think it's really a secret that the moon is the second nicest place out of the earth and the moon. It's a little bit like Antarctica or something. It's not a place that's got resources we really need. What it represents is a symbol, symbol of freedom. It's the place that Neil Armstrong and Buzz Aldrin first tried on back in 1969. And, you know, we came in peace for all mankind. And I think that you and I and NASA and much of the Western world sees the moon the same way. And the Artemis Accords and the Art of Space Treaty and so on. I don't think that the generals in Beijing see the moon quite the same way.

47:18You know, and I might not be quite as hawkish as some of your other guests here. But it does not help me sleep well at night thinking that if the Chinese have a durable and defendable presence on the moon with humans running around there and the United States is not. At some point they could say, well, you know, this outer space treaty is a nice idea, but no one who is alive had anything to do with writing it. And at the end of the day, we have nuclear weapons and this is ours. And then what do you say, right? Like then they basically planted the flag and said, no, the future of the light cone is going to be communism with Chinese characteristics forever.

47:49And that's not a situation, not a future I want. It's not a future, you know, it's not a future or a situation that my children can live up to or participate in. It's not a future that's open to them. They can they can hug the stuffy, you know, rocket to sleep every night, dreaming of a future where they get to explore the universe. But it's just not one that that will be protected, despite the fact that that we spend in the United States, 20 billion dollars a year on NASA and a whole bunch more on other space programs that are less publicized, which is, you know, just NASA's budget is twice the Manhattan Project's budget in inflation adjusted dollars.

48:19It's not it's not a small amount of money. And yet, where's my moon base? Yeah, where's our moon base? I hope the Chinese are taking the proper steps to get Solana's permission to go to the moon. They've got to go through Mike Solana for sure. They've got to go through Mike Solana. What do you hope to see from the next NASA administrator now that Isaac Min is out, which obviously a lot of people in our corner of the Internet were very disappointed by, but I'm curious what you would like to see from them. Yeah, so obviously NASA Administrator serves at the pleasure of the president, and that's his prerogative to decide who wants to serve there.

49:02NASA Administrator is a super tough job, might be the toughest job in government, because you have to balance this impossible set of requirements. And now we've got this budget process that's going on, and there's no NASA Administrator in the chair to be part of that conversation, to be advocating on behalf of which things are perhaps technically sensible or less sensible for reasons that Congress people cannot be expected to understand or even take an interest in. It's a very, very esoteric and strange stuff. And so you end up in a situation where in three months or six months or 18 months or someone, a new NASA administrator will be sitting in the chair with a budget they didn't have anything to do with, a bunch of programs that were designed by people for reasons that are long past their expiry date and have to make the best of what they can do.

49:39And in the meantime, it is 1 ,639 days until China takes the moon. And this is a major, major challenge. I think it's ultimately reflected extremely poorly on the Biden administration that they that they put Senate Administrator Bill Nelson in there, and he was unable to help NASA respond to the existential challenges that face it. His intentions, no doubt, were fine, but his actions and the results of his work were mixed at the best. And I don't think that Trump or anyone who supports him would want a situation like that for NASA looking ahead in four years. And if China's on the moon in 2029, that'll be in the same year that Trump leaves the White House.

50:15So that's a very daunting challenge. This is really a little bit like Sputnikle over again. that we have a situation that really challenges to our core. And of course, Jared is a unique individual. He really, he was not under any confusion there. He knew that going into NASA was a horrible job. Like literally his job is there is to upset a lot of people and get a bunch of programs that are in deep, deep trouble turned around quickly. And it's going to involve a lot of people losing their jobs or getting new jobs, being moved or relocated, all kinds of stuff, just nasty stuff. and the bureaucracy that's entrenched there does not like to change.

50:53And I know people who have worked there. I know people who have tried to change things and have been rewarded for their efforts with constant harassment and even legal and extra-legal problems and so on in their lives, and constant investigations and all kinds of stuff, right? It's just an absolute hornet's nest in many ways. And Jared knew that, and he stepped up because he recognized the United States had given him everything, and he wanted to give back. Now, I don't want to prejudice whoever else steps up. I mean, ultimately, someone's going to have to take the job, and I wish them the best and I'll help them in any way I can.

51:22But it is a tough, tough job. And I think that, frankly, the number of people out there who have the skills to do something useful with NASA and really turn it around and fix the problems that we've got here and the inclination to actually try and to basically blow up their entire life for the rest of their life because that's what's going to result in them at how well you succeed is a very, very small number. It's a real challenge. We have to wonder why is it that in a country of 340 million people, we only really have a handful that could plausibly go in there and solve these problems right now.

51:49And if we don't, China will take our future. Talk to me about power generation on the moon and kind of like the build out if we actually wind up getting there and building like the launch capabilities to get up and back. What does the industrialization process of the moon look like? I know we're going to have to figure out a lot on the fly, but you're in a unique position. Yeah, I looked at a company that there, it was probably a year and a half ago, their pitch was uh we're going to build chevron on the moon okay uh super super bright team that had actually run some pretty meaningful experiments i couldn't quite get there from an investment standpoint because it was like yeah there's so many things that need to happen right but i really hope they're you know successful but yeah break break it down for us well i think step one is is just establish the ability to move stuff to and from the moon right once we have that then we have the operational freedom to build a base of any size we want there right and that's a that's a question for future voters and future presidents, I think.

52:46But ultimately, I'd be quite happy to see something there that's a bit like the Antarctic program. It could even be run by the NSF, right? NASA could be responsible for setting it up and developing the infrastructure and NSF could run it, like the Antarctic program. We send scientists up there because they'll go there voluntarily. It's ultimately there for strategic reasons. And you could have 100 people or 1 ,000 people and you could fly, you know, charconauts for like fly cosmonauts, fly Indian astronauts, fly astronauts from Japan and the EU and Australia and everyone else. And they get to go up there and they work in US moon base run by funded by the United States built by you know American sweat and American brilliance and companies much like the ISS right a little bit like the ISS but maybe with a little bit more kind of down to earth in a way like actually dealing with rocks and and doing real science and stuff like that yeah in terms of energy on the moon there is one one way of getting energy on the moon that won't cost 10 billion dollars to develop and that's to set up large-scale microwave beaming plants on the surface of the earth and beam the power up there oh really yeah the moon always faces the same face towards the Earth.

53:40So once you've got your rectangle panel up there set up, it will always receive the power, provided the power is contingency from Earth. And of course, you could try and set up some kind of nuclear power station or something. But nuclear power plants are extremely expensive to build even on the Earth, let alone on the moon. So I think if you just want to get power there quickly, the best way to do it is just to beam it up. What about just throwing out a ton of solar panels? That feels like, that's what I was teeing you up for. Yeah, yeah, that's what I was teeing up for. Solar panels are great on the Earth because where people live, you never have a night that's longer than about 16 hours.

54:07Sure. and for the vast majority of the world's population. But on the moon, even in the parts of the South Pole, which are ostensibly peaks of a ton of light, you can still have pitch darkness for five or six days at a time in winter, which basically means if you're flying your entire power system up from the Earth, which ultimately you would have to, like 99.9 % of it by mass has to be batteries. And that's an interesting question. You might be better off just flying. I have a starship with a bunch of spare fuel and a gas turbine generator. What about solar panel on one side of the moon, solar panel on the direct opposite side of the moon, really long extension cord?

54:42Very long extension cord. I don't want to bore your readers with discussion of passion's law, but when you run the numbers, it's no cheaper. Oh, that's all good. Okay, so yeah, give me the broad Terraform update. How are things going? And how many solar panels do you have? Where are we in the rollout of solar power dominance across America? Well, that's two separate questions. So at Terraform, we're going great. We're integrating our first Mark I system over the summer right now. It's a machine that's designed to produce limitless quantities of carbon neutral, cheap, synthetic natural gas from sunlight and air.

55:19You can literally roll out a solar panel in anyone's backyard and hey, presto, you're on top of an oil well. It's a very neat trick. Really privileged to get to run this with my friends here and with the confidence of our investors. In terms of solar more generally, look, the outcome is not in doubt. I think Jesse Pelton's been doing a great job on X recently. I've seen him, yeah. Talking about solar, look, at the end of the day, and this is actually another thing where I really hope that Trump and his team come in and help solve a problem that was kind of almost caused accidentally and hasn't been fixed since, which is that just by default, deploying solar arrays is unnecessarily complicated and expensive in the United States.

55:53And really, when we are in a geopolitical kind of struggle right now to see who gets the cheapest energy fastest, and China is far ahead of us, and they're able to produce finished parts for less than the cost of raw materials in the United States, it helps that we're much richer than they are in China. But at the same time, the technology is the same. The sun shines the same. Actually, shines better here. Why is it so difficult to roll out solar panels? It feels much easier than, oh, I don't want a nuclear power plant in my backyard. The windmills, it's going to chop up the birds. There's like so many obvious problems with the other.

56:24Some environmentalists get upset about it, understandably. At the end of the day, having the power to say no is addictive, right? Say no, not in my backyard, even if it's just a solar panel that's making your power cheaper. But actually, the best land for solar in the United States is so hot that no one wanted to live there. The United States government couldn't give it away. You're talking about like Death Valley, basically. I'm talking about like 90 % of Nevada is still federal land. Sure, sure, sure. And so if you want to do work there, you have to kind of do it according to the Fed's rules, which includes NEPA.

56:50which includes like you have to do like four-year environmental impact statements that run to tens of thousands of pages to prove that like putting a solar panel out on a bunch of economically like worthless and biologically lifeless dirt won't harm the environment at the same time as we're subsidizing like 50 million acres of incredibly intense bioethanol corn production it doesn't really make a whole lot of sense no uh so i'm not asking for that to go away what i'm asking is like can we get a categorical exemption for solar deployment with you know maybe some like you can put some money in an escrow account to like take it down in 30 years if you decide you want to go back to being like empty desert.

57:23And that would help a lot because then you wouldn't have to spend, you know, like this is the insane thing. It costs about just the modules itself, the fancy part that turns sunlight into power, costs about 15 cents a watt and the cost of installing it in the United States is about a dollar a watt. So it's seven times the price of the unobtainium to import from China or whatever to install it, which is insane, right? Like if you were talking about the same economics to a corn farmer or something, it'd be like, you are spending how many hundreds of thousands of dollars per acre to install something?

57:50Like, how much does it cost to sow corn? Like 20 bucks an acre? Like almost nothing. That's wild. Talk to me about the supply chain. We hear a lot about the vast majority of solar panels being made in China. Is that a risk? Has it become a political football like everything else, like the electric vehicle batteries, et cetera? Some companies, Apple, Tesla, they've kind of figured it out mostly, gotten around a lot of the tariffs. But what are you seeing in terms of the long-term supply chain dynamics in solar? Yeah, I mean, I think at the end of the day, there'll be countries that make their own solar and countries that don't, which won't really be countries, right?

58:24If you have energy sovereignty, you have sovereignty. And if you don't, you don't. And I think it's very encouraging to see both under, even before the Biden administration, there was strong movements under the Trump one to tariff Chinese solar panels and start the scale-up of solar manufacturing in the United States. And we're seeing like more than a terawatt coming online in the coming years. That's great. At the same time, China is producing solar panels cheaper than anyone has ever even thought possible and they're trying to dump them on our markets And my response to that is someone is if someone is predatory Dumping stuff on you you should buy as much as you can because we can outbid the Chinese solar producers Yeah, and we can just stack the solar panels up in a giant warehouse in Nevada and Until we get around to deploying them But if your enemy is literally trying to give you a machine that prints out money Which is what a solar module is you should just buy as many of them as you can And and deploy them and ensure you can you can prop up the US solar industry you have you want.

59:12It's good to have solar from all the world's factories coming here in the United States to make energy as cheap as possible here so we can make as many things as possible here. So we can do all the aluminum steels producing onshore. We can make as much fresh water as we possibly want. We can terraform Nevada, right? The parts that we don't put under solar panels, we can turn into something like Switzerland if we want to. All this is extremely achievable with cheap solar power. I like the sound of that. Yeah. Can you talk a little bit more about what it actually takes to produce a solar panel in America?

59:39It doesn't seem as complicated as building an iPhone potentially, but maybe it is. I don't think iPhones are even that complicated. We have our intern here. I don't know if you can see. He's making the first USA made iPhone this morning. How's it going, Tyler? Give us the update.

1:00:01I've got the front on, I've got the home button on, I'm feeling really good right now. Feeling really good. Okay, we're making progress. Once you're done with that assembly, TPPN intern, you need to go back to starting with rocks and then make the iPhone from rocks. I've got rocks in my backyard, so I'll give you some rocks. I'll mail it to you. And then you turn that into an iPhone. That's the next level two. Next level two. Yeah. So solar modules, you start with a solar cell and then you laminate it between some glass with aluminum extrusion around the outside. I mean, that's easy. We have factories that do that practically every state in this country right now.

1:00:32If anything's the hard part, it's actually making the metallurgical grade silicon it's kind of an energy intensive refining process um that you can make the the cells themselves out of um but even that is like it's a well understood process people have been doing it for years and years and years the real challenge is like it's like the same thing i wrote in that tweet about like china goes to sleep dreaming about the moon and they wake up angry about the fact they're not on the moon and like it's pretty clear that the nasa managers who are running the moon program are not obsessing over it in quite the same way because if they were we would have a moon base by now um uh and it's the same problem here that like the united states domestic solar industry is not yet at the point where like they can taste eight cents a watt solar modules and they're desperate to produce them right there's not that competition in china there's 20 or 30 different companies like ironically capitalism is working in china to produce the cheapest solar panels ever um and we could there's no reason we can't do that in the states right we're world experts automation we're world experts at capital formation we're world experts at building factories look what tesla's doing tesla's out producing our world electric cars the starlink factory feels like very adjacent yeah the white the white pill is that there's way more complicated than a solar it has to be right it's completely automated yeah it has to be yeah the white pill for american manufacturing is that thousands of people are learning how tesla does it how starlink does it etc and will eventually go and decide hey maybe i should start a factory to make this thing right anyway thank you so much for joining that's it on the fence yeah build a factory build a factory Do it.

1:01:55You heard it here first. Do it. Thank you so much for joining. We will talk to you soon, Casey. Great to see you, Casey. Have a great rest of your day. Have a good one. And the white pill on sales tax automation. Of course, it's numeralhq.com. Spend less than five minutes per month on sales tax compliance benchmark Series A. Anyway, next up, we have Patrick McGee, the author of Apple in China, a groundbreaking book on Apple's supply chain. Very excited to have him in the studio. Welcome to the show, Patrick. How are you doing? I'm doing great, but I haven't watched WWDC. So I hope you don't have any questions about that.

1:02:25No, we don't need to talk about that. The timeline is not loving it. I have a post here. Why does Apple have a whole conference just to show everyone they slightly adjusted the colors on a few icons? So that is a little bit dramatic. But in other news, we have our intern here in the studio assembling an iPhone from scratch. Not really from scratch, but I don't know if you can see. But he's getting it together. he's assembling the first iPhone in America. And so I think that's more of our focus today. Oh, God, I said iPhone manufacturing would never happen in America. I think this is the first American-made device.

1:03:05You're witnessing history. He's starting, it's a 4S. So we couldn't go all the way to, you know, whatever the 16 or the 17, whatever we're on now. But we had to start somewhere. But anyway, thank you so much for joining. Would you mind introducing yourself in the book and kind of like what led you to want to dig into this so much? Because I know you interviewed like hundreds of people for this. It was obviously a massive project. Take me through the journey to actually start investigating this story. So, yeah, Steve Jobs would sometimes say you can't understand your life until you connect the dots backwards, right?

1:03:39It'd be a real trick if we could know where the meaning was going forward. That would improve things. So I've spent 12 years at the Financial Times, and it was basically like three years in Hong Kong, where I was getting to know China, Three years in Germany, where I was covering the automotive industry, but in retrospect, really understanding supply chains and manufacturing, industrial policy to some degree. And then four years covering Apple. So you sort of put those three things together and really get my book. The first three years of covering Apple, I did what most Apple reporters do, which is that I didn't focus at all on how the sausage gets made, as it were.

1:04:09I focused on product features. I focused on the rise of digital advertising and Apple's privacy policies. And I think I had some decent stories. but it didn't really occur to me to look at what we all know, which is that the products are made in China. And the real discovery that I made, if you will, is that we had never covered a job within Apple called manufacturing design engineer. And these are the people who go to over to China, literally by the plane load to train, audit, supervisor, and mount billions of dollars worth of machinery in the aggregate in hundreds of factories across China. So I wasn't really familiar with what makes Apple most unique in hardware, which is that they don't sit around hoping that component makers in their supply chain come up with more durable materials, cheaper materials, lighter materials, et cetera.

1:04:53They go and they train all of the people to come up with those innovations. So it's a co-invention process, really. But the amount of people they've trained and the amount of money they've invested in China to make all that happen was staggering to such a degree that I compare it in a way that sounds unhinged if you haven't sort of read the 200 preceding pages in the book to the Marshall Plan. Yeah. Can you give us more on the exact scale of the investment and how you actually pulled those numbers? Are they available in SEC reports? Or did you have to do some digging to actually piece everything together to actually get the full scope of the scale of what's happening?

1:05:27Yeah. So the line I have on this is that when Tim Cook goes to Zhang Nenghai, the citadel of communist power, and invests or rather pledges to invest$275 billion dollars into Chinese factories over the next five years if he were to hold a press release a press conference I should say from Zhang Nenghai that would have been corporate suicide this was when Donald Trump was campaigning on America first platform and so was that 2016 2016 yeah so basically what happens is that when Xi Jinping comes into power particularly when he becomes president in March 2013 Apple fears that its products are going to be blacklisted because there is a three-week digital blitzkrieg against the company where all sorts of state-sponsored media are threatening the company, and even regulators are saying, threatening things to Apple.

1:06:15And Apple basically doesn't understand. They're in a state of confusion. There's not a single senior person living in the country, and there never has been at the time. And Apple responds by basically hiring or naming a team of people that call themselves the Gang of Eight, and they are to be Cupertino's eyes and ears on the ground. Essentially, they are tasked with coming up with a way to placate local and federal officials in China without resorting to joint ventures, because that is anathema to Cupertino. A joint venture is like if the three of us wanted to go make a company in China, 50 % of it traditionally needs to be Chinese owned.

1:06:44That's supposed to be illegal after 2001 when China enters the World Trade Organization, but it really just evolves rather than is prohibited. And so Apple has been operating under the radar to such a degree that even though by 2012, they are the largest producer in China without actually producing anything themselves, You could sort of compare it to Uber in that respect. They're also the largest retail seller in the country among foreign countries, foreign companies. And so they're wildly successful, but they're operating so secretively that the Beijing leadership doesn't understand what contributions they are making.

1:07:17And so Apple does its own supply chain study to sort of flip this on its head. And what they realize is that they're investing$55 billion a year into the country. And so they sort of take this to Zhang Nenghai to basically like say, get off of our back. You have no idea how much we're contributing to your companies. And if I'm just around this out, I would say if we fast forward to today, 55 % market share in global smartphones is the Chinese manufacturers like Huawei and Oppo and Vivo and so forth. And the reason they're so good is that Apple trained all of their suppliers. Yes. I remember you making this point about how Apple is essentially like the reason that Oppo and Huawei exist because they trained all of the manufacturing houses to build competitors to iPhones, essentially.

1:08:05And you made the point that Apple didn't kill Nokia with necessarily like product innovation in the United States. It was the manufacturing and the birth of the Chinese manufacturing supply chain that was taken advantage of by Huawei, Oppo, and the other Chinese phone companies. But is there a reason why Nokia wasn't able to go to China and say, hey, we're going to piggyback on the same supply chain? We're okay being a year behind on the lagging edge. Did you ever dig into the exact dynamics by that? Was that just a blunder by the Nokia team? Because it seems like if you see this rising manufacturing powerhouse, you might want to go and take advantage of it.

1:08:43if Apple's gonna pay for all the R &D. Yeah, and it's the same question from Microsoft with the Windows phone, right? Totally, totally. The trouble is going to China is not the secret. If you've watched the BlackBerry movie, in fact, this is the final scene where they do have the BlackBerry Storm, I think, just made in China. Yeah, yeah, yeah. It's a really simplistic way of thinking about it because it's not just that China has the skills and capabilities of pulling this off, it's that Apple has this pyramid structure in the company where Johnny Ive and the design studio called ID were given like a godlike power to come up with unfathomably intricate and difficult designs.

1:09:17And then the manufacturing engineers basically had to go make it happen. Nobody had the skills to go make it happen. And so the line that I got from one of these manufacturing design engineers is that the perfect attitude was to say to Johnny Ive, I have no fucking idea how to build this, but I'm going to move heaven and earth to go figure out how to do it. And nobody but the Chinese were willing to sort of put in the investments, the tailor-made industrial policies, build factories on a scale that was needed to build iPhones at the volumes that Apple needed. So China just emerges as the perfect answer to Apple.

1:09:50Before we get more into what happens next, can you talk about some of Apple's early experiments? I know they were trying to do stuff in California, the Czech Republic. It sounds like they didn't. I know. And so, yeah, there's this seven-year narrative, and it sounds long, but trust me, it's fast paced and it sort of, you know, crosses continents where Apple doesn't just sort of like, you know, abandon making its own computers, right? Because they used to make them on three continents and turn to China. That's a seven year experimentation stage where the translucent iMac alone was built by LG in Korea.

1:10:23And then they expanded to building them in Mexico and in Wales. It's a great strategy. Unfortunately, it's actually a fiasco. The Mexico factory is literally on fire for about a month. And so production has halted. And the production line in Wales was known as the toaster line for the propensity of the circuit boards to catch fire so they have all these other problems that just like make China begin to look better what happens is Foxconn comes on board so my line about this is that Johnny Ive and Steve Jobs the meeting of their minds is what makes Apple products unique but it's the meeting of the minds between Terry Guo founder of Foxconn and Tim Cook that makes Apple products ubiquitous so when Foxconn comes on board this is a Taiwanese company that's true yeah and I was gonna and I was gonna say did they feel more comfortable partnering with Taiwan?

1:11:06So they begin with Taiwan. In the first five years of Steve Jobs' comeback, Taiwan is the most important country. It's where a lot of the products are being made, namely the iPod, but also the PowerBooks and the iBook. If you remember Reese Witherspoon in her little bunny ears in Legally Blind, lining up to buy a computer, she definitely buys the little bright orange computer. I mean, Apple is redefining aesthetics, not just of computers, but later of MP3 players and later, of course, of smartphones. And so I like to joke with people, like what's your favorite Dell computer from the early 2000s, right?

1:11:34Because nobody has anything, you can't even remember any of them. Nobody was doing the sort of thing that Johnny Ive was doing. But what we've forgotten and what my narrative sort of fills is just like to have that maniacal obsessiveness in design is one thing. And that's where a lot of the stories are. But to actually build those in the 10s of millions per year, and later the hundreds of billions per year, like clearly, there's a story there, right? And it's weird that we've omitted it. But to build that quality, that quantity, and that cost, China just emerges as the perfect partner. So it's the combination of Cupertino's skill and China's scale that makes for the perfect marriage.

1:12:09Yeah. What's the stat on the Tim Cook era? It's something like 20 % of smartphone sales, but 80 % of profits, something like that. Isn't that insane? And I'm only giving a figure for 2016, but it's actually increased since then. So the market share for the iPhone has never gone above 20%, but these days they reap 85 % of industry profits. Wow. That is just insane. I mean, find me another sector that comes anywhere close to that. It's remarkable. Yeah. Give it up for big tech profits. Yeah. How much of what you're seeing today with, there's some criticism of Apple for kind of faltering on the AI side, on the product side.

1:12:48Maybe the iPhone's gotten a little bit stale with innovation. And to be clear, Mark Gurman just said they didn't really announce any new AI features. Yeah. Which surprised him. At the same time, while people have been going after Apple for the lack of product innovation, lack of software innovation, it feels like in this particular moment, Tim Cook is the perfect CEO for Apple in the sense that they are facing immense geopolitical challenges. The supply chain is more complicated than ever. And Tim Cook seems to be a fantastic leader in terms of the execution of that strategy. And so we're in this weird, I understand that there's a weird dynamic, but it feels like in terms of the challenges that Apple, the size of the challenges, they could miss on AI.

1:13:32But if they're able to scale up to a billion iPhones a year, they win. Hard disagree. Okay, break it down. I understand the point that you're making, and Tim Cook has provided immense value to Apple, but I do think it's sort of a time's up moment. The reason I say that is Apple is so wedded to China that it has basically become an existential risk to the company. I'm not saying therefore the company is going to downfall. But if something like their export license from China was canceled, it's not like, oh, well, thank God we have 25 % of iPhones now being made in India. None of those phones are independently made in India.

1:14:06They're all reliant on the China-centric supply chain. And if Apple is going to come out of it, I mean, basically, if you agree that that's an untenable situation, it's a 15-year effort if it can happen at all. And you can't have the guy who created the current strategy leading the company, if that's the goal. So I would just question that he's the right person. And that's just on the supply chain side. That's on the operations side. That's his wheelhouse, if you will. The AI strategy is probably just as fundamental in the sense that we're now at a stage where people's next phones could very well be based on what AI features they have.

1:14:36And Siri, to quote Sachin Adela, is dumb as a rock. And they're basically not even allowing you to have ChatGPT replace Siri. I mean, ChatGPT is not allowed to go through my emails, not allowed to go through my photos, etc. It could probably work wonders if it could, but Apple, for privacy reasons that may or may not be genuine, can't do that. And when you get into China, Chai Chiang Kaiji is not allowed in China. And so AI is really taking off with the likes of Baidu and Alibaba and DeepSeek, of course. But Apple is going to be put in this strange position where, you know, my book is about how they've trained up their hardware engineers in China over the last 25 years.

1:15:10Is the next 25 years the story of how Apple is training up the AI developments in China? I mean, for starters, I don't know that they're a great teacher, just given that their own Siri product and AI isn't all that good. But that strikes me as an untenable position for them to be in. And so AI is like another, maybe not existential risk, but certainly a large risk for Apple that they certainly are flailing. And if the buck stops somewhere, it's with Tim Cook. So I would say there's two reasons that you'd call for new leadership right now. So talk about getting out of China, building up manufacturing capacity in India, Vietnam, America, anywhere else.

1:15:45Tim Cook has done it once in China. He doesn't have any unique tie to China. He's not Chinese. He just went over there, figured it out, built it up. It took immense time and investment. He's done it once. Can he do it again in a different country? Is that possible? So you sometimes hear that within Apple, and I would say that it reeks of hubris. In other words, China was a once-in-a-century partner in terms of the tailor-made policies, in terms of literally having hundreds of millions of people go from the migrant labor pools and into cities like Shenzhen and Suzhou to make all this happen. So it's not that China was just some blank slate with a lot of people and you could get it working.

1:16:20It's not just the density of the population, it's the dynamism of the population and the way that they're willing to work and the sort of bonded zones that China has. I mean, there's so many reasons why none of this is going to happen in america because even on a philosophical basis we wouldn't want to replicate things like the dual class uh structure of citizenship that you have in china where if you're rural you really can live in shenzhen but you can't raise your kids there you can't get the social benefits you certainly can't retire there you know philosophically we don't want to replicate that uh whether india has to go that far if they're going to do so i don't know but they at least are a country of 1.5 billion people they have wages that are similar to the early 2000s in china and they do have major policies of wanting to become a manufacturing powerhouse.

1:16:59I don't think they're doing enough, but it's not just a matter of Tim Cook sort of copying and pasting the strategy into a new area that doesn't take into account just how different India and China are. Can you tell us more about that difference? India's democracy, but there's a lot of lack of infrastructure I've heard flagged as just like there's not enough scale at the ports. There's not enough industrial manufacturing equipment in the country. Same thing with Vietnam, although it does feel like Apple's maybe had a little bit more luck with AirPods in Vietnam. But I don't know if there's more color to that.

1:17:32And it might be more of one of those like final assembly things that we've been hearing. On the whole, I would say more is actually happening in Vietnam than people give credit for. And less is happening in India than people think. The problem with Vietnam is you just hit constraints, right? I mean, the population in Vietnam is literally one-fifteenth that of China. So it is never going to rival China in a way that India could. India just doesn't have the skill. I don't know that they have the drive. They don't have the same Nietzschean will to power. The epigraph of my book is from a Made in China 2025 document from 10 years ago, which says without manufacturing, there is no country, there is no nation.

1:18:08No other country, let alone India, thinks of manufacturing in that way. But China does. So it's really difficult for India to compete. And even though they have a lot of people, they don't have the same migratory patterns. So I studied a study that looked at maybe 80, maybe 84 countries of internal migration, and India was at the end of the list. So you don't have this culture, especially of young women, 17, 18 years old, leaving their parents' homes and going to work in a factory in Karnataka as like an equivalent of Shenzhen. Now, maybe you don't need to replicate every facet of how China pulled this off for the last 25 years, but it's not as though I look at India and see a bunch of qualities that would help them in manufacturing that China doesn't have.

1:18:49It's sort of one way that the advantages that China has. And as the world moves into automation, it's a little disconcerting that I believe I'm accurate in saying China has more industrial robots than the rest of the world combined. And they certainly have more than an order of magnitude more than in India. A lot of the issues around moving Apple's manufacturing around the globe today come down to demographics like you're outlining and just cultural differences. How does Apple's leadership think about automation? Because if Tim Cook is willing to invest 50 to hundreds of billions of dollars a year in a country, isn't there a world in the future where Apple's investing$100 billion into American iPhone manufacturing that's end-to-end automated?

1:19:35I know that's extremely difficult challenge. It kind of gets around the cultural issue. But it gets around the cultural issues and it just becomes a capital formation issue. The stat that we were toying with was that I think over the last decade, Apple has returned almost a trillion dollars to shareholders via dividends. Yeah. And so that's enough money to do really, really big mega projects. The question is, is it just even that's not enough money? No amount of money can get you there? Or is there some other factor? Yeah, because I think when people say like, oh, it's impossible to automate iPhone production.

1:20:09I don't think that the average person genuinely believes that in the fullness of the next 20 years, there's no way an iPhone could basically, you could put the parts in one side of a factory and a finished iPhone would come out the other side. So let me say two things. One, I'm totally in favor of stiffing shareholders and cutting the buyback program in half and spending the other$50 billion a year on building resiliency into the supply chain. So I'm not exactly sure that's what you're suggesting, but I would advocate that. Secondly, even if, let's just say, artificial intelligence on steroids is able to upend the way things are manufactured and you don't need 400 ,000 people in Zhengzhou to do it, nevertheless, China has, unfortunately, a number of advantages in terms of the refining and the mining of the metals.

1:20:51Even if the metals are being mined in Africa, it's Chinese companies that are orchestrating all of these processes. So even in a state where it is, as you say, it just sort of lights out factory where you throw in the components and then iPhones come out, that's probably still more likely to happen in China than America, because they're the ones who have all the expertise sort of end to end in the supply chain. And then, you know, shipping an iPhone, even on a flight to US is immaterial in terms of the costs. So you really would need the sort of political catalyst that we're seeing now in order to make that happen.

1:21:22But it has to sort of be the perfect convergence of everything happening. And Tim Cook can't just be the only person who's being pressured by the White House to do this. Because if you're going to have Foxconn and a whole bunch of industrial clusters setting up shop, you know, in Pittsburgh or wherever it is that Trump wants it, they can't just be supplying iPhones, right? There is no factory in the world that all they do is supply iPhones, right? Think of a TSMC building chips for any number of company that you can think of. That is what you need. So it's not a matter of convincing Tim Cook to open a factory.

1:21:49There's no such thing as an Apple factory. You need Tim Cook to convince his underlings in the supply chain to build things. But in order to do so, they need to be building for, I don't know, Sony and Samsung and a whole host of industrial clusters are so a whole host of companies in an industrial cluster. Otherwise, the economics and the efficiency don't make any sense. Yeah, I remember you mentioning that it sounded like Apple had a mandate that their suppliers would not be overly reliant on Apple to the tune of more than 50 % of what they produced being for Apple. Why did that dictate come from Apple?

1:22:19You would think it would come the other direction where if I'm a supplier, I don't want to be too dependent on Apple because if they pull out, I lose my entire business. That's true, but Apple has all the leverage in these relationships. And so if Apple says jump, the supplier says how high. And so if Apple's really, and just think of how demanding Apple is and particularly how much they scaled, right? So 5 million iPhones were built in 2007, nearly a quarter billion by 2015. So if you were a supplier just trying to keep up with that, like that was enough business. Even if you were making 1 % or 2 % margins, the volumes were making you grow at a level that you had never grown before.

1:22:53But yeah, what happens is that if Apple obviates the need for a certain component and turns direction just as they go from 100 million iPhones to 170 million iPhones, well, you would just go out of business. And so this would happen, and it would develop bad will from the supplier community towards Apple. And so Apple began saying for its own protection, however fast you grow with us, please grow as fast with somebody else. And that's what sort of gives birth inadvertently to the Chinese smartphone industry. I don't mean the suppliers. I mean Huawei, Apple, Evo, etc. So speaking of the consumer market in China, how cooked is Apple in China?

1:23:26I know sales are down double digits as of at least last year. It doesn't look good, especially with all these new AI features coming online. And is that driven by just consumer choice and nationalism? Or are there actual structural policy choices that are driving that change? My question back to you is whether cooked was a pun. Always. We make tons of cooked puns. I just wasn't sure how intentional that was. Oh, yeah, of course. Yeah, I mean, well, let me give the sympathetic views first, which is that after Trump 1.0 kneecapped Huawei. Huawei was at the time, first of all, outselling Apple globally in terms of smartphones.

1:24:06But more importantly, it was the only phone to penetrate the top tier. Apple doesn't really care if other phone makers sell more phones than them, but they do care if they're in the top tier. And so when Chinese luxury buyers were purchasing Huawei phones instead of Apple phones, it was a really big deal. But Trump disallowed Huawei from having access to Google services, which maybe doesn't matter so much in China. But I mean, I used to live in Germany and had a Huawei. If that thing didn't have YouTube and Gmail anymore, gone. I'm not going to have it. And they also disallowed them from having 5G technology from Qualcomm.

1:24:41So that worked really well at a time. And China market share for Apple was 9%. And then when Huawei gets kneecapped, all the luxury buyers in China that need a 5G phone, just as 5G was really coming on the scene, they all bought iPhones. So the market share doubles to 17%. You could say that as a result of that, as a result of Huawei being on the sidelines, Apple's market share was sort of overly inflated. So now that Huawei is back, we have had two years, and now this is the third year of Apple market share declining in China. But if I'm to be sympathetic, I would say it might just be that they're sort of right sizing or like refining an equilibrium now that Huawei is back in the game.

1:25:18Unfortunately, I wouldn't sort of go with the sympathetic view. I'm more cynical here because now that Huawei is back and you're seeing more nationalist fervor in China and now that iPhones are going to be built in India, which actually diminishes the sense that the iPhone is an indigenous made product in China. and now that AI is taking on more preeminence in terms of why you should buy a phone and Apple's having all sorts of problems on their own and the problems are only compounded by the fact that they can't use partners like ChatGPT in China, I would predict that by 2030 at the latest, iPhone share in China is gonna fall back to single digits.

1:25:52Interesting. Can you talk more about the parallels or differences between Huawei, Xiaomi and Apple? Huawei's gone into GPU accelerators, they're competing with NVIDIA on the Ascend side. Xiaomi has a car. Apple's kind of stumbled in cars, or there was rumors that they were working on a car. They didn't. I guess my question is, why hasn't Apple reaped some of the downstream dividends of adjacent product development based on their supply chain dominance? Because that seems like almost free money for them. And it seems like all the Chinese companies have been like, oh yeah, we make phones, So let's also make hoverboards and cars and everything.

1:26:35And there's been a much different culture in terms of product development. Does it just come down to culture? Is there something else or kind of structural going on? It's a fair question. I don't really know why Apple abandoned the car project. And I think it could have been a winning product, especially in China, because they have such brand appeal. On the other hand, I would say the Apple car product probably should have died years earlier. The reason I say that is because Apple got caught up in the hoopla that Silicon Valley was in in 2015, which is that self-driving cars are going to be everywhere, right?

1:27:05I mean, the founder of Lyft had said that, quote, a majority of our rides will be self-driving by, I think, 2021. Travis Kalanick had said that they would buy 100 ,000 Mercedes self-driving cars by 2020. And Larry Page of Google said self-driving cars will be bigger than Google. I mean, there was a sense that self-driving cars were the next big thing, and Apple was not immune, and Apple fit into it. As soon as we realized that the self-driving thing was a bit of a fantasy and it's going to take much, much longer to develop, I think the project could have died. Because at that point, it's just like, is Apple just competing on acceleration with Tesla?

1:27:41Like, what exactly are they going to do? If you're in a self-driving car and you're using iOS apps on all these, like, you know, sexy windows and stuff within the car, like, it makes all kinds of sense. Once you take the self-driving car element out of it, I don't know that the Apple car makes any sense. The Chinese are in a different ecosystem where it does make sense for them to have a Xiaomi car, a Huawei car, et cetera. And maybe they're self-driving, maybe there's not. but they sort of like having an advance on all this technology and they like market share. Apple only wants to be in markets where it's going to be a number one or number two player.

1:28:09It doesn't want to play second, third fiddle to Tesla and any number of other companies. So it was kind of remarkable that they actually went so heavy into Apple TV because they were behind HBO and Hulu and Netflix and they really like carved something out there that worked. But I know a lot of Hollywood people that were very skeptical about that because in film and television, the brand standard is so high and you have a couple bad shows, you get a bad reputation, but they slog through. Can I give you my theory on why they moved into Apple TV Plus? Please, yeah, that'd be great. Apple has a services division that is largely not what you think it is as the consumer.

1:28:43In other words, if I said, what are Apple services? You would say, oh, it's Apple Music. It's the random$11 charge I get a couple times. What it really is, is 30 % fees on App Store and Google paying them$20 billion a year. So for the purposes of earnings calls, I think they want to push more into things like talking about Ted Lasso and what Emmy Awards they won, rather than having to say, we squeezed more out of Spotify this year. And by the way, Google is now paying us$22 billion a year. I really think that was the motivation for why they went into TV+. Why do you, somewhat random, but why do you think they never made a television?

1:29:17Oh, I don't know. I mean, I've heard one thing, which is that Steve Jobs demanded, this is obviously before 2011, that when the TV was off, that the pixels be white rather than black. And that Apple was unable to nail this. I'm not saying that's why the project died. I mean, it's so interesting because I would happily never get any Apple media products. I don't need their shows. I would much prefer to have an Apple device instead of this buggy Samsung. you know absolutely but it's funny tvs are good right like when apple comes up with the ipod it's because they're looking at mp3 player on the market in 2000 2001 and they're saying everything sucks the reason why they never made a video cam even though they came out with iMovie is that video camcorders were generally speaking doing fine so they needed to break into one market yeah but i would argue i would argue that tvs suck i have i hate every i've got a mess i hate the four TVs in my house.

1:30:13That's a Samsung frame in the back there. I think that thing's great. It's okay. I have one too. Yeah, but the software, it's buggy, it's slow. It wants me to upgrade and subscribe and do all this junk and it's just like nuts. I guess the question is how much you would pay because TVs aren't a high margin. 10 grand. Yeah, I mean, I think it would be like the Apple Pro Display XDR which is like a$4 ,000 screen. Yeah, it's basically more than more than anything else based on the specs but it's still so good that designers still demand it And I think that they could pull that out. But I think you're probably correct overall.

1:30:46Last question, and I want to let you go. I want to hear the story of the mini Tiananmen that happened in 2010. That's a fascinating story. Oh, God, what a terrible story to end on. It is the only bit in the story that I would call salacious, and I only have it in there. I don't like salacious stories, is that it's just an incident that takes place where Apple really realizes they're in over their head, and they don't understand the culture. How many seconds do I have to tell the story? A minute or two. Two minutes. Okay, so it's hard not to, like I need to give the backstory, which is that the iPhone 4 in 2010 goes absolutely wild in China.

1:31:23It becomes the most conspicuous thing that you can own. And it's so conspicuous and there's such a supply-demand imbalance that Apple at the time has four stores in a country of 1.4 billion people. And so these scalpers known as yellow cows begin sending migrants to the store by the thousands where everybody can each buy two units. And then these gangsters who are backed by the triads are basically like lining up these people, getting all the phones that come out of the store and then going to a city like Chongqing, population 32 million, number of Apple stores zero. And they're able to sell them at seven, eight, nine hundred dollars.

1:31:57They begin finding ways illegal, legal and legal to make more money per iPhone than Apple. And one of the ways was to go buy a phone in America, T-Mobile, on a 24-month contract, and using a fake ID, they could acquire the phone for one month worth of the 24-month contract, smuggle it into China, and then sell it for$800 or$900. I mean, that's just the state of demand at the time. More than that, they were actually zapping out the phones. So they were rendering the thing inoperable. But in the process of breaking it, they were masking the fact that it was a stolen phone and then making more money that way as well.

1:32:29and they were even hollowing out the phone's components, selling off the memory and things like that. So if you understand just like through this brief thing that there's this chaotic scenes of crazy demand of triad gangsters and everything, one of the worst things that happens in the book is that Apple runs out of product at the Shanghai Pudong store. So there's 7 ,000 migrants waiting to buy their product and they're more scared of the triads who have hired them to be there than they are of the police. So when Apple says we're out of product, you need to go home, they don't want to leave they feel like they can't leave and they literally prevent the glass doors from being shut on them and they stay to sit in in the store so the police are unable to get them to go the mayor of pudong shows up they're not able to go and eventually after 11 p.m a hundred black cat black clad police officers that someone compared to like you know the uh the swat team or or to uh what's it called steel team six um they those kind of guys show up and when a young woman sort of having never seen these people takes a photo of them one of the persons takes her behind the genus bar and beets are so bloody that the blood on the ground cannot be scrubbed out and they have to re-import granite from Italy to replace the stone tiles it's a terrible story the person who told me it has been having nightmares about it for the last 14 years but it's an example of Apple just not understanding what they were dealing with and it results in the political awakening the following a couple years when they begin to hire senior people on the ground to understand the politics the culture and the distribution issues that they're having in the country very rough well uh it's been fantastic talking to you i personally thank you for for apple to reshore manufacturing and um diversify the supply chain i'm on the side of tim cook i think he can do it i think he's cooking i'm on the side of friend shoring i would love to see more in mexico and more in india i don't know what's going to happen in ditzburg yeah Patrick, thank you for doing the work to tell the story and come back on anytime.

1:34:25Thanks much. Okay, cheers. Talk to you soon. Bye. Really quickly, let me tell you about Adio, customer relationship magic. Adio is the AI native CRM that builds, scales, and grows your company to the next level. And we are power users. Every day. Power users over here. We are power users. Next up, we have Aaron Slodov coming in the studio to tell us about Reindustrialized, all the things that are going on in his world. I'll be right back. And let's check in on the intern cam, see how Tyler's doing. Aaron, how are you doing?

1:34:59At the donut. Oh, yeah? Oh, looking good. I like the background. We have our intern in the studio today making the first iPhone in America. It's happening. It looks like he's done. Give me the update. He's getting pretty close. I'm just saying, most of the external things here, I got the volume buttons, home button. I got to put in the battery, the main chip motherboard. Okay. But feeling pretty good. I'm probably 80 % of the way there. 80 % of the way there, less than an hour and a half in. I think this is going fantastically. I wasn't sure if this is possible, but it looks like it's happening.

1:35:36Anyway, Aaron, how are you doing? What are you up to these days? uh doing well and getting ready for uh you know the second annual industrialized conference give me the background on the first one how'd that go when and where is the second one yeah um so this is kind of a labor of love of many people right uh basically you know we have we have a nice team of co-founders myself austin bishop phantom donahue and greg bernstein um and an amazing team of organizers but yeah last year the backstory is kind of obviously you know hard tech deep tech the wave was kind of happening at that point um i put out that article the techno industrial manifesto that's right lots of questions lots of people talking about this across military government finance whatever um wanted to put all these people in the same room together right so So the idea behind this was, can we actually build a big tent around this idea?

1:36:40And I saw it as a way to really just like convene. Who were some of the top speakers last time? So we had Sean from Palantir. We had Eric Prince. We had the lieutenant governor of Michigan. I mean, it was crazy, right? Like this all came together in under three months, basically. And we were all grinding super hard to bring this awesome group of people together. Chris gave a keynote from Hadrian. And it was this really, really excellent group of people. It was like you were there with a thousand of your friends, right? And everybody's ready to build. And it was wonderful in so many ways. And people really thought that it was like, yeah.

1:37:27What is the overall state of the reindustrialization project? There's been a lot of thought pieces and conferences. What have we seen on the legislation side, on the EO side, and what are we waiting for next? I mean, yeah, we could probably spend an entire hour on all the things that have rolled out of the admin since January. But it's been surprising, to say the least. Like, wow, that's awesome. What's up, Aaron? Good to see you. Hello, you as well. So yeah, following up on last year, one of the things that we established at the very end of the summit was we wanted to bring together the new industrial alliance, new American Industrial Alliance, NIA.

1:38:12So Austin has been running that in D.C. for the last year, and we've got basically a mountain of legislation and policy that has been active and pumped out through various EOs. we have reconciliation happening right now, there's going to be a ton, right? And shipbuilding standalone by itself, right? I think we're going to be at a level where the sovereign wealth fund potentially, right, like equipment financing out the wazoo, like it will be pretty interesting to kind of like watch and see what happens here. Very, very excited to see. What can you tell us about what to expect at this year's re-industrialized?

1:38:56Are tickets on sale yet? Where can people go to sign up? Is it invite only? How is the conference coming together? Any speakers you can tease? Yeah. So we did actually bring out a little list the other day of our top invites. And you can check that out on the re-industrialized account, which is re-end summit on X. But yeah, basically the way we're doing this right now is that it's mostly invite only. We're balancing, you know, a very interesting community here in terms of like bringing the stakeholders together, decision makers, but also all the people that are very enthusiastic about reindustrialization and bringing manufacturing back.

1:39:43So getting Aaron's DMs. That's what I'm hearing. Start yapping. That's great. And I mean, big, big development since the last re-industrialized. I've seen the iOS autocorrect keyboard stops correcting re-industrialized to de-industrialized. Total cultural victory. Let's go. Let's go. We love it. We love it. I think that's subtle signaling from Tim that we're bringing the iPhone back to America. Yeah. Okay. Give me. Any advice, Aaron, for Tyler? I was about to ask. Yeah, yeah, yeah. What's the advice for our intern who's assembling the iPhone in America for the first time? What tools does he need?

1:40:24You know, what are the key what are the key steps to really reshore manufacturing? I mean, there's hardly pieces that didn't go in that phone. Yeah, take it apart. Take it apart. Put it back together. Yeah. Yeah. Yeah. Aaron, what does it take to reshore the iPhone? Is he, I mean, are we going to give him a gong ring for this? Because I mean, it's a lot of, it's a lot of grit, right? We're watching history. So cultural shift is what I'm hearing. Yes. I mean, massive cultural shift, which is, you know, underlied by a lot of other factors too. So I think that infusing and transforming the industrial base through technology gives, you know, a lot of incentive for people to jump back in and join.

1:41:05We're not, we're not trying to go back to the 50s here so um i i really do think that that's you know part of the cultural movement uh right to like bring the stuff kicking and screaming out of uh world war ii era or the 80s or however you really want to think about it into the 21st century and it's not going to be easy right like listening to patrick talk about that stuff is it's mind-bending honestly like how how much one company can actually do right at scale um so yeah yeah what are you thinking about kind of the the first steps to uh biggest re-industrial is like what is the rate limiting like in the supply chain right now we've heard about mining and dirac's working on work instructions and hadrian and you are working on different pieces of uh whether it's cnc or injection, there's a whole bunch of different pieces of the subtle supply chain.

1:42:00I talked to one person who was working on bringing a team internationally to build brushless motors for drones, for small drones, and they were like, oh yeah, I'm excited to get to America, set up shop here. I'd love to just set up my company in the brushless motor district in America. Wherever the brushless motor district is, I'll just set up my business there. So then if I need a certain part, I'll just walk across the street to another company. And the guy to explain like, no, no, no, we don't even have a district. We don't even have a company. We have nothing. So yeah, what are you most excited about?

1:42:31What's perhaps underrated right now in terms of the supply chain? So it's a really good question because this is kind of like, can we pull the Uno Reverso card on China or whatever other supply chains and bring some of the stuff back? What does that actually require? And it's more than just the raw materials and machines. it's the way that I think about this stuff it's like trade skill right like these are very very precious pools of skill that when you give them away right it empowers somebody else to scale them somewhere and if you've got hundreds of millions of people to do that that's it's pretty challenging you know we don't necessarily need to outdo them we just have to be right like in the arena still.

1:43:18Because one of the things that they want is to de-industrialize everybody else, right? So we can't, we just can't let that happen. And that's okay, right? Like we have to be able to fight back on a fundamental level by preserving, replicating, and like scaling this very precious like trade skill. And if that has to be done through AI and automation, right? Like to compress the stuff that people really shouldn't be doing, right? Like these dirty, dark, dangerous types of jobs. We don't really necessarily need to go back to that. But yeah, there's a lot, right? Like underlying this problem. And I think a lot of people have a challenging time separating the fact that like, do we actually need to be focused on making t-shirts and toasters in America?

1:44:06But like, not necessarily, but the feeder industries that you need to do all that stuff, you know, grant you the ability to do all these other things, right? Yeah, yeah. I have this line about like maybe we need to make Happy Meal toys in America in order to earn the right to make iPhones in America too. The supply chain is so interlinked that even these kind of like these silly toys that seem so easy to offshore, maybe we need to bring those back too. Anyway, it's been fantastic having you. I'm sure we'll learn a lot more at Reindustrialize. Where can people go to learn more? What's the website?

1:44:37Reindustrialize.com. There we go. That's a great domain name. Let's give it up for a great domain. Thank you, Aaron. Thanks for hopping on the show. We will talk to you soon. Have a great rest of your day. And we're excited to cover it. Thanks, Aaron. Before we bring in our next guest, let me tell you about public.com, investing for those who take it seriously. They got multi-asset investing, industry-leading yields, and they're trusted by millions, folks. Our next guest is from Soligen. Soligen is Soligen. Oh, we'll ask him. Welcome to the show. How are you doing? Would you mind introducing yourself so we can get the pronunciation correct so we don't mess it up?

1:45:10We hate messing up pronunciations on this show. How are you doing? Good. It's Solugin. Solugin. Okay. You want to know how we got it? Yeah, please. It's a stupid name. Hydrogen peroxide. Solugal hydrogen peroxide generators. There we go. It rolls off the tongue. You know, the portmanteau era in Silicon Valley, deeply underrated. There was a time when everyone had a portmanteau name. Then we went to the LY name. You could have been Solugin Lee. Clulee's bringing it back. Clulee's bringing it back. You had Bitly and all these different names. I think there was a.sy, like Syrian domain name for a while.

1:45:46Oh, yeah, that was a big one. A lot of it was driven by like what country is open to selling their TLD. Notion, yeah, Notion. Mostly the war-torn countries. Yeah, yeah. Yeah, Notion inspired the.so domain, which is Somalia. And then I remember.tv was Tuvalu, which is a tiny, tiny island nation of like 5 ,000 people. I might be getting that wrong, but I'm pretty sure at one point they were making like half of their GDP off of selling.TV domain names. Yeah, so.AI domains represent 20 % of the government's total revenue for Anguilla. Anguilla, wow. There you go. That's fascinating. Anyway, we're not here to talk domains.

1:46:26We're here to talk industrialization, making massive things, chemicals. Can you give us a brief overview of the company and the shape of your business? and then I have a bunch of questions about how you actually like build this stuff. Absolutely. So, uh, solugin pronounced solugin, uh, we got started in 2017, uh, spun out of, uh, MIT, uh, and then went through Y Combinator, um, raised a significant amount of capital to basically decentralized chemicals production. If you look at chemicals production today, it's like these massive facilities. I grew up in Houston. My backyard was effectively a refinery.

1:47:01These are the size of islands, like Galveston is an island. It's largely a refinery. What we said was like, what if we can marry my background and my co-founder's background, Sean, to basically make a better process? And we did. We invented basically, I guess you can call it the world's smallest chemical factories is how we think about it. And the whole point of that is that instead of having these massive centralized facilities, you've got decentralized production where you can put these facilities anywhere in the world. and actually have your product be closer to the chemical. We started the business with hydrogen peroxide.

1:47:36Very dangerous to ship hydrogen peroxide. So the whole premise of the company and the name, as we discussed, is basically creating peroxide on demand near the end customer. We've since expanded quite a bit out of not just peroxide, but added multiple chemicals into our portfolio. So, yeah, that's kind of the background. I mean, I'm familiar with using hydrogen peroxide. if I get a little cut out of my hand, like I imagine that that's not the vast majority of customers you're talking to? Yeah, no, it's a good question because when we started, everyone was like, you mean like the little brown bottles on the shelves?

1:48:10We're like, yes. And there's a huge market for peroxide for water treatment. So it's a really good antimicrobial. So it kills bacteria, but also for rockets. So like if you look at like rocket propellant, a lot of it can contain high concentration peroxide. It's called HTP peroxide that basically gives you pretty good thrust. It's like liquid oxygen, but more powerful. And then the other applications are like in paper and pulp and all the things that no one talks about. It's like this invisible multibillion dollar industry that we look after. Wow. I mean, you said the plants are smaller, like not an entire island, but I imagine that it's not something I can put on my desk.

1:48:48How big are we talking? Yeah, no. Unfortunately, your intern cannot make one of these at his standing desk. not yet i don't know i don't know look at him look at him this guy tyler can do anything the first american iphone i think he's doing he's making he's making the first usa made iphone he's doing well he had to take it back apart to put the battery in oh the battery's going in now it's looking good the battery's in he's doing great uh yeah so how big are these plants yes it's like 10 000 square feet so basically the size of like a small warehouse okay great yeah So that's the idea. Got it.

1:49:24And then walk me through the chemical supply chain in America. Are there, I mean, we hear a ton about, we don't have enough tool and dye makers in America. If we're talking about Apple, we don't have enough rare earth mineral extraction. There's this deja vu. Anytime you find out about a new thing in the supply chain, it's always bad news and America's not leading in it. Maybe give me the bad news and we can end with some good news. um i'm not going to shock you guys here 80 of chemicals that we consume in the u.s come from china okay sorry they're not booing you they're just they're just patriots you know it makes sense though because it's the cheapest cost of capital and the cheapest cost of labor out there so what we're saying is we want to go after chemicals that give the united states an edge that China can't make easily, that are difficult to ship from other countries beyond China.

1:50:23And the United States is uniquely capable because of the technology we're developing to produce. So if you look at the supply chain today, it's a producer, basically a giant chemical factory, makes the chemical, concentrates it into a dangerous mix of something that can fit into a tanker truck. That then goes to a middleman who then sells that to another middleman and then to another middleman. There's about five middlemen in this chain. They're called distributors who then sell that to the end user. What we've done at Solusion is said, hey, like, what if we kind of collapse this whole middleman chain and put one of these units near our sites?

1:51:01I'm actually I'm pointing to my plant. We have a plant behind me right now. That plant, our biggest customer is 10 minutes down the road. They take most of all even the plant. And basically that's the premise of what we're trying to go after. and the business. Yeah, I mean, talk to me about the early go-to-sales motion because like selling chemicals does not feel like, oh yeah, you're a YC company. You're just going to sell to the other companies in your batch. Yeah, we sell a lot of other things actually. No, I'm kidding. We'll buy a million dollars of your CRM. You buy a million dollars of hydrogen peroxide.

1:51:32We both have a million dollars in ARR. You're really breaking the YC narrative. I mean, now YC obviously does a lot more hard tech. They do a lot of cool stuff. stuff but you were you were early to that party for sure we were early to that party and the way we went to market um it's kind of a bizarre story so we started in um what was called float spas have you guys heard these these are like these isolation things yeah that are like 25 weight percent salt and you float in it and you're completely isolated from from the world turns out they use a lot of peroxide to clean that water oh that makes sense but when we started we weren't at scale so we We couldn't make industrial quantities of product.

1:52:09We can make small artisanal batches of peroxide. That's how we kind of pitched it. But then if you look at these float spas, they're not using that much, but they were paying a lot per gallon for this product. So we just said, OK, let's just go a little bit cheaper than what everyone else is doing and sell it to the float spas. We got 80 percent of the market share within two years. Congratulations. Not a big market, but we were able to kind of like dominate it. That's the YC. that's the yc that's the founders fund thing you know that's our entree into that world and then from there once we had that win uh we of course went to consumer products we actually launched a brand which really like not many people know the story we actually uh launched a brand called ode to clean so basically we said okay we did this whole float spa thing we sold off um basically we had a partner that we just got a contract with and sold on the volume then we said what's the next what's the next highest leverage point we can get as we scale the technology?

1:53:06And it was wipes because everybody wipes. And so if you have these wipes that have just a little bit of peroxide in it, you have a pretty high dollar per gallon of peroxide. So we launched this brand, a cleaning wipes brand called Ode to Clean. That got bought out by the nation's biggest wipes manufacturer. We got a supply contract out of that and that really kicked off our B2B sales version. Yeah, that's amazing. Now, get me up to speed on the fundraising side. I'm interested to know you raise all the capital, but the use of capital has to be different because there's CapEx in the business. There's R &D.

1:53:45But talk to me about the mix of venture and debt and how you think about deploying capital, not getting over your skis because you're raising big rounds, probably ahead of huge profits. How are you underwriting these investments as they happen? Yeah, that's a great question. I think fundamentally we look at this metric called return on invested capital or ROIC. That is like our North Star, where any asset that we go after has to pass an ROIC filter. And so we only go after the top two ROIC assets at any given time at Solusion. So fundamentally where we are today, we've raised about 700 in growth in venture capital.

1:54:22That's fantastic. And then we've also we got a Department of Energy loan program offer for$214 million at 6%. But that program is in this administration a little bit challenged just because it's a Biden era thing. No big deal. We're going to private banks to curb the debt side of it. Now that we've proven out commercial scale, we can basically show that this is the unit economics on one of these assets. This is how big it can get. Let's go after multiple funds. Yeah, I've talked to a couple of hard tech investors or industrial founders, hard tech founders, where when they have to build something, there's a whole bunch of different structures.

1:55:04Like, do they want to own the land? Do they even want to own the building that they're building? Or do they want to set up a different LLC for that and raise different investors for that? Have you gotten into that game? Yeah, we've done everything. A lot of people say if you're doing project financing, you should put it into a project financing entity. Yep. What we fundamentally believe is like we can move a lot faster if there's no one else kind of involved at the table. That's going to dictate how fast we can move. Traditionally, like I would even say that even the LPO, they wanted us to work with like this white shoe engineering firm that would have taken three years to build our factory when we did ours in 11 months.

1:55:41So we said we're not we don't really want to work with those folks. We're going to do it ourselves. And so that's why we own everything outright. Got it. Talk to me about automation and just getting more value out of the capital assets that you're deploying. Humanoids are in the news. That seems pretty far off. But what else is valuable to you right now in terms of are there things that you can design to make the plants more efficient? I wish I could sit here and tell you like the automation piece is going to be big for us. It's not. I'm like not for what we're doing. the fundamentals of chemicals is all about yield and yield comes down to chemistry so if you look at a traditional petrochemical process the yield is about 60 which means 40 of your initial carbon is either a waste product or has to go into the environment as co2 what we're saying is let's actually engineer that first before we try to optimize or automate anything because that's the tree bottleneck for the cycle time of these products so that's where we're focusing on the yield side.

1:56:45I will say there are opportunities for automation that are popping up that are exciting. That's on the defense work that we're doing. That is going to be quite important because we are making products that will go boom. And it's important to have a lot of automation and control around that. Yeah. I imagine that that's more. Yeah. I mean, I'd love to hear more about that to the extent that you can talk about it, but also in terms of any of the advances on technology in terms of understanding chemistry. We've seen alpha fold and there's amazing things that you can do when you have a lot of data.

1:57:20Is that something that's coming down the pipe or is that just not really? We actually just released a paper that got published last week called, our algorithm is called Badass. So basically what it does is it basically figures out a sequence for an enzyme. So we use enzymes as catalyst. It figures out a sequence that's most likely to have the function that you're going after. And that paper showed 75 % predictability based on the sequence itself. That in itself is important, but what's more important is really the reactor design itself. How we think about how the chemical's interacting. It's truly an engineering problem versus a science fair problem.

1:57:57Yeah. Jordan, anything else? If you're successful, is there a specific vintage car that you would develop? If you know Enios, the Enios Grenadier. Oh, yeah, another industrial chemical manufacturer. NEO's fourth largest chemical manufacturer. They've redesigned and rebuilt the Land Rover Defender. Yeah, the Land Rover Defender. Is there a car you would look to? A Camry or a Subaru? That's how he started. A Subaru. That's great. We're rebuilding a 2005 Camry. It was the perfect car. It was the perfect car. That's amazing. Thank you so much for hopping on. Awesome. This was great. Thanks for jumping on.

1:58:33Cheers. Really quickly, let me tell you about AdQuick. AdQuick.com, out-of-home advertising. made easier, easy and measurable. Say goodbye to the headaches of out-of-home advertising. Only AdQuick combines technology, out-of-home expertise, and data to enable efficient seamless ad buying across the globe. If you're an AI company, go spend$100 million with AdQuick right now. Right now. I think that's the best thing you could do to differentiate yourself. Anthropic, if you want to break through, you want to break through the noise, get some consumer downloads on Claude. Billboards. Every billboard.

1:59:07Every billboard. Not just a few, every billboard. Let's do it. You have the capital, deploy it, get aggressive. Speaking of capital, we have venture capitalist Sean Liu from Founders Fund in the studio. Welcome to the stream. How are you doing, Sean? I'm doing well. Let's give it up. I'm great. Good to see you. We just talked to Garab from Solugen. Is it Solugen? I keep messing it up. Solugen. Solugen. Solugen, yes. I want to hear your side of the story. How did you initially wind up working there? What did you learn? and then we'll go into some of your venture career. So I'll start with how I ended up there.

1:59:44It was a reference from NAP and Brian St. I'm kind of like a crazy tech guy that likes capital expenditures. Any company that has tech risk, capital expenditures, bring it on. Let's give it up with crazy capital expenditures. The more dollars, the better, right? I love it. And I was thinking about getting my hands dirty with something early, something fun, and I saw that you came up, met the guy, love the vision, love the team, and it was a hard job. We were small in the middle of COVID and needs help. And I feel like CFO, COO is one of those underappreciated roles. So I took a chance on myself and the company and it's been a lot of fun.

2:00:21Second question, what did I learn? I think probably the... It's all management theory is bullshit. Any book you read is complete bullshit. You go in and you learn on the job and whatever advice they give you is all fake news. yeah i mean what what's an example of that what was the first moment where you were like oh this this completely i gotta throw out the playbook on this one you know i think there's there's this whole thing about you gotta be nice to your employees it's just like hey the hr gurus and the team is gonna tell you you know you can't work your employees after 5 p.m and you're just like we're a fucking startup let's go right like you work on saturdays you run you run you run night shifts, there's something called a DuPont schedule for a reason.

2:01:12You run 24-7 and factories don't go down just because you need to take a break. So yeah, I think it's just setting those cultural guidelines. And sometimes when you hire people, you hear people say, hey, you got to define your culture as where we have all these missions and values, which is true. And at the end of the day for startup, do you have the IQ, the grit and the grind to suffer through a marathon that's a startup. So it's almost like a different profile. How much of the work at that time was focused on fundraising and kind of educating venture capitalists about underwriting what is kind of a first-of-its-kind company to really go through the Silicon Valley pipeline versus kind of the operational role of the CFO, COO, actually deploying the capital?

2:02:03I would say it's both your firefighting. every day, every minute. I remember when we joined, there was no one in the company to do finance. And we had a really bad head of FP, and I fired him after two weeks. So I went to account inventory myself. And, you know, we're not a SaaS company. You got to account inventory. You got to know what - That's like barrels. Your warehouse. So it's important. You got to know what every piece of the equipment costs. And on the flip side of that, yes, There is a huge, you know, capital makes a difference, right? Cost of capital. We haven't had a new chemical startup that's scaled in, you know, more than 30, 40 years.

2:02:44There's a reason for that. Partially it's because there's a tech barrier where R &D is centralized among these petrol giants and they don't really want to release IP. And the second piece of this is cost of capital for startups is so high. You build a factory, probably costs you, you know, 300 million bucks on the low end. We have a revolutionary technology. you still got to bridge through the valley of death. And for hard tech, capital intensive business, you have more of them than you typically see in SaaS. So educating them on how to think about risk and them, I mean, venture capital is how to think about risk.

2:03:20At what point should you see inflection in value? And what does that actually transcend into is, I would say, is the other half of the battle. Yeah. On the IP side, uh are you like were you watching for i for patents to expire and then implement them immediately are there other ways around these problems like like what was the ip landscape at the time are there still like trade secrets that are just uh important that you could just focus on maybe reverse engineering or finding a different pathway like what what was that like at the at the company at the time i mean i think you know there it's really interesting right like i think patents only get you so far.

2:04:00That's why, you know, if we have faith in patents, China wouldn't be where it is. But I think the key is what, I would say if we take a big step back, a lot of the formulation, though, how it still sits with the petrogiants, they have a lock in, and they have a lot of resources to continue to be patent trolls. I remember one of the, you know, probably two months of the job, you know, right about when we were about to close the round, one of our competitors sent to assist and assist to us that was like, oh, you stole our formulation and it's like tiny little variants in something that's complete bullshit and they can still sue you out of oblivion, right?

2:04:39But I think we've kind of learned over time to in some ways you kind of you do have to fight some of the battles similar to what they do. You file all these umbrella patents as general as possible to protect your core IP and on the other side you kind of have a lot of trade secret that's very difficult to duplicate. But at the core, because how different Soligen's approach is to thinking about chemistry and how to scale this, right? It's a generalized enzymatic based platform that, you know, it makes that side of it much easier for us to navigate. Talk about the transition to Founders Fund. How'd you wind up there?

2:05:14And then I want to go into some of the deals that you've underwritten and how you think about some of what's going on in the hard tech landscape broadly. Yeah. Um, it was similarly, you know, I, I was, you know, my, my family is in the Bay area. You don't want to start a family. And I was thinking about what are the places where I get to work with the smartest people building and investing in the craziest things and still be able to help solid and a founder. Transparency was only twice. I couldn't see myself working in any other VC firm. I don't think it fit very well on any other VC firm. And, I'm very grateful that the guys took me under their umbrella.

2:05:49So obviously you've looked at a lot of hard tech companies. At the same time, valuations feel very high. Walk me through how you're actually underwriting a growth deal in a somewhat capital-intensive business. How is it different than looking at a software company? What are the major pitfalls or landmines that you need to be looking at when you're building an investment thesis? and is there anything that's kind of like underrated that most companies or VCs might be missing when they're underwriting these hard tech companies in this like kind of new era yeah man it's a it's a broad question you asked and I think it's a hard one to answer because within hard tech and capital intensity those are two different things right sure last year you know Napa me and I did a deal in Crusoe and it's more asset and intensive there's technology involved but the way we think about something like a Crusoe, which is more of an energy thesis and long-term bet on, you know, the infrastructure shifts that's happening in the U.S., the energy gaps that we see is quite different than what you typically see in something like a Neuralink or SpaceX, right?

2:06:59Where if we take a step back, if you think about what is the technical risk you face, and the first question we always ask is, do we need to invent science? And if you do, then that's a very different question. And very transparent, Foundersign is in the business of backing revolutionary technologies and companies. The end part is important. So how does this science transfer into a company is sometimes missing in those discussions. You see really brilliant technologists building these cool tech and you realize that if you believe you need five, six, seven new fields of science to all come together and have that to be true, then we are probably not the right place to be.

2:07:40But then as you kind of progress towards the true growth stage, are we thinking about scale at risk? If so, what are we dealing with? Is it like no unknowns or unknown unknowns? And what are the different ways of thinking about that? Sometimes when you look at, let's say, a new generation technology, a lot of the questions we say, look, the science is proven. We don't need to invent science. Let's look at what does it take for this to materialize in such a way that we as investors transparently can actually generate a return that makes sense in a time frame that makes sense. And a lot of that hinges on do you have the supply chain ready, the upstream fuel sources, turbine generators and everything else that's associated with building something like a new generation technology?

2:08:22do you believe that if you figure this shit out, can the rest of the supply chain also figure this out with you? And a lot of the time the answer is no. And that makes it that much harder. And the last thing is I do feel like in most hard tech spaces, founders have a tendency to oversell the technology. And they have a deep founded belief that the moment, you know, the technology is mature, people will come. And that is not true. Right. And I think a lot of the questions I think Trey, Peter, Knapp, and others ask on the team is, right, like, how are you going to sell this thing? Just because you make it doesn't mean people want to buy it.

2:08:57And I think some of the best investments we've made are, you know, in hard tech are the technology companies that doesn't need to invent science. They're making incremental improvements to scale. They understand the bottlenecks of the supply chain, have a realistic plan to deal with the bottlenecks they see. And then lastly, the market is already there, right? If you think about SpaceX launches a market, yes, it's expanded, but they have a really strong anchor to kind of take their to take market share very quickly so that commercial mindedness on day one is super important yeah it's the same thing with starlink where people are already buying internet and then they just switch their provider or add on a secondary provider uh jordi you have a question do you see hard tech as a category becoming something that goes the way of maybe the creator economy was in 2020 where a lot of you know maybe there's opportunities, but a lot of investors get so burned that it becomes kind of this flash moment in time?

2:09:50Or is it something that you see the average venture capitalist continuing to invest in consistently over the next 10 years? How skewed is the power law? Because I mean, if you look at the history of like dev tools or like databases, it feels like there's like 10 companies that produce venture returns in most of those categories because a bunch of them got out in IPO, much of them hit multi-billion valuations. It felt less than your take all. Yeah. There was a point in time where people were saying, I'm going to start investing a lot of money in software as a service. And then those people, if they were decent investors, 10 years later, they still said, I'm still a software as a service investor.

2:10:28The question I have is every VC over the last two years decided, I'm also a hard tech investor. And I rationalizing it by saying you don't need to be technical to invest in hard tech. And there's definitely evidence that that can be true. But I'm curious if you think the average venture capitalist, even three, four years from now, will still be deploying into broadly hard tech. You know, history doesn't repeat, but it rhymes. I feel like people don't learn and will continue to do that. I think, you know, what's interesting is we probably see a more, I would say hard tech as a category is probably is a misnomer, I think you'll see more finely parsed views of hard tech, you are using bio, are you, you know, biotech pharma, you biotech infra, right?

2:11:22Like, I think that that should, that we shouldn't look at hard tech as a category in its totality, right? I think defense tech historically was grouped in as hard tech, if you look at this, you know, 10 years ago right and rule is hard tech today it's its own category um and i do feel like people continue to to to invest right like the nature of technology is hard tech very transparent all new technology that emerges hard tech so you'll continue to see that but the moment you create one power log company you see you know the whole host of siblings and children come out and then it becomes its own category and people invest in it you'll continue to see this you know cycle of life and death Symbio was super hot in 2020, 2021.

2:12:04And I feel like that's now a taboo term. But at the same time, you feel like there's tremendous amount of progress in technology. And I feel like a lot of the company we meet on the AI for science side is pretty much, I would say, the next generation from a fundamental tech perspective, addressing what biology can do. Right. You see a lot of funding going into it. But to reframe your question another way, how do you see hard tech generating returns for venture capitalists? I think hard tech, more than software is de facto power law, right? There's one SpaceX. That's, you know, 99 % of the market.

2:12:38I mean, space is not a market space in many ways is SpaceX, but there's a lot of different derivatives that opened up. You know, Android today captures most of the mining share for defense, and you see that, you know, in other parts of the hard tech ecosystem. There's very few pharma companies, and they're massive in comparison. And I think the reason why I think this is more defensible than software is because you do have a huge amount of capital advantage and cost of capital is a huge advantage here. When you're Dow DuPont, you have hundreds of billions of dollars of CapEx on the ground. The marginal cost to produce is so low that it's very, very hard for incumbent to come in and displace you.

2:13:19And that's why we feel like there's fewer power law companies here. But if you bet on the right ones and double down on the right ones, you can create massive venture return. Do you have a take or like a post-mortem on what happened when venture capitalists went really big into clean tech? I think this was like the late 2000s, early 2010s. My kind of naive take on it was that it was well-intentioned based on environmental concerns in many ways. But the category definition was too narrow. and so they didn't include Tesla as a clean tech company. Tesla, obviously, if you had included it in the clean tech bucket, would have been the power law winner and driven massive venture returns.

2:14:04But instead, Tesla was seen as a car company and as a result wound up not being included in that. And so all the clean tech funds underperformed just the general software funds. In terms of category definition and just post-mortem on the clean tech investment thesis that we went through a decade ago. What's your takeaway from that era? Yeah. You know, Gaurav had this really good saying where he was like, you know, if you're one step too early, you're a genius, two steps too early, you're a martyr. And I feel like clean tech 1.0 was the martyr category. It's kind of like building Dropbox in the 1990s.

2:14:47You didn't have enough, you know, bandwidth wasn't high enough. Internet wasn't quite a thing. And in cleantech 1.0, the industrial base is not quite there, right? Polysilicon processing capabilities was not there. And, you know, the utilities' ability to manage non-form systems' ability is also not there. So there's a whole host of factors that kind of made the cost structure was not competitive. manufacturing scale up was not there. You need to invent new science to make this work. And you also need to then do sales and marketing to companies to consumers, as this actually made sense. And you're also coupling that with selling to utilities, which has capped ROIs.

2:15:29So, fundamentally, a bad customer was bad setup all around. And Kugan, to your point, I think the pitch of a narrower definition of which is purely focused on generation technology is also, I think, misplaced. Tesla is not a generation technology, it is a car. You group it with car technology that shifts it. And I think if you think about nuclear today, it is clean technology. I mean, yes, we're rebranding it, but nuclear is the cleanest of all technologies. So I think this whole wave of how do we think about creating and broadening that definition so that you see where the clearest commercial pull is coming from is probably more important than the definition of what the technology is or is not.

2:16:16On the flip side of that, too, I think the last generation was highly dependent on government subsidy. Then, no, like if you look at underwriting, whether it's from the DOE or others, like all of them assume a certain level of government subsidy that lasted for decades on it, and that just was not going to work. And I think that's important to, you know, as we think about investing in hard tech and, you know, a lot of these new technologies, when does the subsidy go away? At what point are you competitive? Yeah, that reminds me of the story of Solyndra. heavily like a picked winner by the government.

2:16:47And it feels like the Solyndra example in Zero to One at least is kind of like, I don't know, I call it like a spreadsheet error. It's like they built their spreadsheet for the economics of that business and they just missed one variable. And it feels like in hard tech, there's a number of companies that are in that category today where it sounds really good in a pitch deck, but once you actually run the economics, If there's one number that's off by some small factor, you wind up with completely upside down economics. And so in terms of underwriting hard tech companies, how important is it to go through the long term economics?

2:17:26Do you need to bring in experts or run the first principles analyses on these companies? I'm just seeing a lot of companies that are like, we're gonna do this on the moon or this in space, And I'm just like, oh, if launch costs don't fall by 10x, like you've assumed you're completely upside down. That seems like an important part of underwriting. Or is that usually less of a factor because you're focused on the growth stage? It's still very important, right? Like I think when we think about that, the question we always debate at Founders on is what is, how many like need to beliefs do you have to make the thesis work?

2:18:00and if you need to believe you know if you believe a and b and c by the point like this codependency yep you multiply them you get zero right and i think of this very much as like a semiconductor manufacturing process 7 000 steps each step has 0.1 percent yield loss by the end of it you get no chips out so so it's it's it's similar like do you do you need to believe you know to your point if you let's take like you know some crazy thing like asteroid mining in you alpha centauri do you believe you need to create like nuclear fusion engines launch costs you know robotics and then you're like maybe i just back all these companies and then wait for that to happen right so that's that that is a very very important thing i think the more and statements there is the worse it is and very transparent we don't like to make those bets right like the cleanest companies are the ones with one or two of these and very much it's you're you don't need to invent science with very little dependencies um you know when we look at general matter when look at soligen or or or others in our portfolio you do see that you know in your link even for that matter it's you you don't see that many and statements a lot of those have been proven out and what we need to understand is really like at what point like what do you need to believe for that to scale um and um and i think i think to your point on like how do we do due diligence i mean yes i think it's it's that we do need to understand science there's a zero one question Like, do we take on binary risk and where does that exist?

2:19:27And then there is a, you know, going back to Founders on Roots, what's our read on the founder? And how do we see them in tackling all the risks coming up? Yeah, and the execution side. Well, thank you so much for joining. This was a fantastic conversation. We'd love to have you back. Enjoy the rest of your day. And thanks for breaking it down for us. Love how candidly we are. Always fun. Cheers. We'll talk to you soon. Cheers. Bye. Let's check in on Tyler in the interim cam. He's got the glasses on. You know that means he's towards final assembly. Come back to me in maybe 15, 20 minutes. I'm almost done.

2:20:01Okay, almost done. He's cooking. That's great. He is cooking. I think he has another maybe hour in the show. It looks like he's making progress. We'll check in with him at the next break. First, let me talk to you about 8 Sleep. Get the Pod 5 Ultra. How did you sleep last night, Jordy? Use code TBPN. If you're going to get one, they're fantastic. Five-year warranty. 30-night risk-free trial. risk uh free returns free shipping i got a 91 how'd you do jordy i'm sick so oh oh excuses excuses excuses oh john beats you one time and all of a sudden the excuses come out what'd you do what'd you do give us the number sorry the the wi-fi is not working yeah okay anyway number is horrid 63 let's hear it for me that's maybe the worst i feel like i i come out strong in the beginning of the week.

2:20:49I usually beat you Monday, Tuesday, but then you catch up and then you just go on a run. Anyway, get an AIDSleep. Go to AIDSleep.com slash TBPN. And we will bring in our next guest, Garth from Panthalossa, building a very cool technology. I'll let him tell you about it. How are you doing? What's going on? Hey, guys. Great to meet you. Thanks so much for having me on the show. I love the show. I'm really excited to have you. I have heard about your company. The background. What's going on here? Is that real? Yeah, this is real. We're in Portland right on the river wow that's beautiful it almost looks tropical yeah the lighting looks yeah yeah very cool and you have the plant inside the window so that adds to the ambience looking great uh would you mind a quick introduction on you and your and the company and then we'll go into it yeah yeah so company's called panthalossa we're based in portland about 80 people a lot of people out of places like spacex uh blue origin virgin orbit it's it's a deep tech company love the conversation with Sean just now.

2:21:45And, you know, the premise of the company is to go to the middle of the ocean and capture energy there very cheaply. So this was an idea that got started nine years ago. I was at Bridgewater Associates doing investment stuff. But before that, I had been doing energy stuff at MIT and working on a startup for energy. And my co-founder had been also looking at the ocean and we were saying to ourselves you know what if you could find an untapped energy source on the planet uh bigger than many of the ones that people have been going for and uh and maybe more scalable maybe lower cost of capture so the one that really stuck out to us was the open ocean far from shore where there is wind all the time and the wind makes the waves and so you can be out there, uh, capturing hypothetically, what we thought back then was at a very low cost.

2:22:42And, and then you would, you know, use the energy right on site for things. Um, and so this was sort of an era before, um, Crusoe, but maybe a similar idea that you go to where the energy is. And, and we thought that this was really the best place to go the middle of the ocean. And so of course, of course you need to develop a tech stack for that, but there's lots of interesting properties there that you can talk about. And what we thought was that maybe you could build a really simple system. Like, you know, a lot of the stuff that's been done in ocean energy before was like pretty complicated and pretty expensive.

2:23:17So we thought that if you could develop a really simple system, then you'd start to get a lot of those properties. I'm hearing simple and nine years. It sounds like those are some famous last words. You've been grinding on this for a long time. But it sounds like there's actually breakthroughs. I saw some photos. You've actually built one of these things. Can you explain how big it is, how much energy it captures, what you're deploying, what the rollout schedule is? Because it's been a lot of work to get to where you are today. And I want to hear about like, where we are before we go into where you're going.

2:23:50Yeah. Yeah. I actually think that that length of time is a really important part of any company that's trying to do deep tech. Totally. You know, you want to stay small and under the radar while you're sorting everything out. Right. I think that a mistake that a lot of companies make is that they go to market with their first idea or their second idea. And so if you can arrange things so that you can throw things out a lot before you settle on a path, because the fact is you are going to get it wrong. You're going to need to iterate. And particularly when you're trying to do something where the cost is the key consideration, or if it's space, maybe it's weight.

2:24:29You want to give yourself that time. So when I look back on the first four years of that, which was really the heavy R &D time, I'm extremely glad that we took that time to hone the technology and get it exactly right. And so by the time we got to year five and six, when we were deploying ocean systems, we were right on the money in terms of power. We were hitting all of our simulations right away. And it comes from the simplicity of the system where what we decided was that you want a system that is no moving parts at all, except for one small water turbine. So these systems, we call them nodes.

2:25:04They're about 10 up to 20 meters across at the top. And they go down into the water column about 80 meters, maybe 100 meters for full scale ones. So big systems, but pretty small in the scale of things like shipbuilding. So this is sort of the shape of one of them here it's a you know and it sits in the water like this and it goes up and down with the waves and all of the dynamics are coming from the inertia of water within that system and as the system goes up and down it pumps the water into a reservoir just like a hydroelectric dam and then from there it drives a water turbine so so it's a really simple method of operation and once you have that you can simulate it effectively and it's really easy to mass produce as well.

2:25:49So you keep saying, you keep saying simple and I'm looking at your website and it's like the most complicated thing I've ever seen. But I mean, it speaks to, it speaks to your background and, and, and how ambitious this project is, but I keep thinking you're, you're really, this is simple stuff, John. Yeah. Yeah. It's simple. Get with it. Just, just, just weld and build this massive thing. And there's all this wiring and we're pulling up your website right now and it's fantastic. I mean, congratulations. It's amazing. Um, I, I want to know more about the applications, how much energy are we generating with one of these, and then I want to go into some of the applications because it feels like you could not have possibly predicted the reinforcement learning revolution that's happening right now, the ability to do distributed AI training, and it is the perfect time for this thing, and maybe you just made your own luck, but I want to know about the evolution of the application, the strategy energy thesis, and kind of the scale of what we're thinking about doing here.

2:26:47Yeah. Yeah. Great. So each one of these systems makes anywhere from 200 kilowatts to a megawatt. So you can build anywhere in that scale. Got it. And, you know, when we got going, it was interesting because, you know, obviously we were jettisoning the cables back to shore and the moorings and all of that stuff. And we said to ourselves, you know, this was back in 2017 or so. What are the applications? for which this can be very effective. And we decided on two platforms. One is computing, and that's the primary and first platform. And then the other is making synthetic fuels and getting the fuels back to shore.

2:27:26But if you look at our patents and everything back from 2017, 2018, we have a whole bunch of them on computing.

2:27:38And back then we were thinking about AI. you know, often in the patents, we would call things brain simulation. So we had AI brain simulation, crypto. Sure. Probably Bitcoin at a certain point. Crusoe was doing the same thing. Yeah, exactly. And, you know, I had been at Bridgewater. I'd been doing AI stuff with Dave Ferrucci there. You could see even at that time where this was going. You could see it in the exponential growth of energy. You could see it in the growing power of the models even then. You know, this is when AlphaGo was happening. So, you know, you could sort of put two and two together.

2:28:05And what we believe, of course, is that, you know, this is going to 100 gigawatt scale, terawatt scale. And there's not that many options on the planet for making that go really fast, at least if you're not doing something like coal in China. So we are positioning ourselves to be an option that can scale really rapidly, much more rapidly than terrestrial infrastructure. and that's really the whole play. We want it to be more sustainable, more affordable. It's much cheaper and be able to scale to much larger levels even than most terrestrial infrastructure could. Got it. So you said up to one megawatt, is that right?

2:28:48Yeah, yes. The nominal system might be 400 kilowatts, but you can go larger than that. So how big of a, are you thinking like it's basically a data center inside? Like you could put a whole bunch of H200s, maybe get up to like a few thousand maybe. Is that roughly correct? Yeah, exactly. Like a few hundred up to a thousand GPU equivalents. And there's enough room in that device to actually run. Yeah, I mean, that's amazing. So what are you tracking in the reinforcement learning training market? Is that important because you're at this point where you're actually going to market? and will it even, like, are there specific decisions that you need to make on the equipment that you load onto these machines or are you trying to serve the GPU capacity or data center capacity at, like, a higher level of abstraction?

2:29:43Exactly. So those are the sort of questions about the go-to-market. But as you identified, there's, you know, the massive push towards test time compute, just tons of long running inference. Yep. It didn't take workflows all the time. Combined with, yeah, I mean, we think it's going to be mostly RL on the training side, you know, and I mean, I'm sure you had people on last week and you saw the semi-analysis thing drop. Yep, it was great. Yesterday. But yeah, I mean, you could sort of put it together even when you had Alpha go back in the day of like, you put together LLMs with reinforcement learning, you get very powerful models.

2:30:18So the future is largely about who can deploy the most RL. And then once people really start to use this, you know, we think it's going to move to all kinds of applications beyond text. You know, this has to move to engineering. It has to move to medicine. And you're going to want to have agents working together around the clock. So you have long running inference. And basically what this all comes down to is you want the cheapest throughput for tokens. And so what we're targeting is that long tail of very massive horizontal scaling of distributed energy units. And that's what we think our platform can be.

2:31:00Over time, would somebody live on one of these systems? Would they visit it frequently? We had Dylan Patel on last week, and he said one of the challenges with putting data centers in space is that chips are unreliable. Sometimes you need to like plug, you know, basically unplug a system, plug it back in. It's the IT meme. Unplug it and plug it back in. Basically, you know, even with cheap energy, the IT work will still be necessary. So I'm wondering, is one person going to be responsible for a number of these systems and they sort of visit them and make sure they're operating efficiently? There's probably other solutions to full tolerance.

2:31:37And I can imagine you're building some type of gyroscopic tech to kind of keep the system stable when they're shifting up and down. I think that doesn't even matter. I don't know. Yeah, but I'm curious. It's really important. So this is part of the tech. So you not only need a way to produce the power on board really simply and cheaply, you also need to be able to drive them around. So these are self-propelled systems. And they're self-propelled without things like propellers. So they create their own thrust just because of their shape. So we've done trials where we're driving these things in figure eights out there.

2:32:12And it's all happening just because of the shape of the system. Wow. And what that allows you to do is now this is sort of like SpaceX boosters coming back. You can deploy the systems relatively close to shore. They work their way out into the resource. They hang out for a period of time capturing energy. And, you know, maybe it's three years, maybe it's five years. When it's time for a payload swap, they bring themselves back in and then you swap the payload. That's remarkable. So you can configure that. Yeah, you can configure the payload swap time for whatever you need to get the numbers, you know, for the meantime between failure on your payloads.

2:32:45And at the beginning, I'm sure we'll start with the most reliable stuff. And then as we get into worlds of inference ASICs and stuff like that, that are, you know, more reliable than the GPUs, then we start deploying those as well. Yeah, that makes sense. Are there lessons that you're learning from the Bitcoin ASIC era, like other kind of out-of-the-box, stranded energy projects? Crusoe was kind of widely known for the, I think it was gas beaker plants initially, but obviously those are very much terrestrial. You can just drive the ATV or the truck out to them to check on them. What have you seen or learned from the more creative energy solutions to the Bitcoin ASIC problem?

2:33:33Well, I mean, I think the big theme, of course, is everyone shifting towards AI energy. Just because the amount of demand that's coming is enormous. Bitcoin is, you know, Bitcoin is always going to be there as a fallback for anyone who's got very cheap energy. On board our systems will have between two and three cents per kilowatt hour, probably less in the future. So that's always an option for us. But I think that as people look at this, it becomes clear that the model layer will become more commoditized. I think there's a lot of comparative advantage there for a long time. chip layer too. But the fundamental bottleneck is going to be who can scale tens of gigawatts per year or even 50 gigawatts per year.

2:34:20That's a really hard thing to do in the West. Obviously, we've been doing net zero gigawatts per year for a long time. So if we don't relax that constraint, then you get prices going up and that becomes a major impediment to the build out. I mean, so the device already is massive. If we're adding a gigawatt, you're talking about making a thousand of these. Walk me through the actual manufacturing, how you scale that up to go from, you know, kind of one exquisite system to making dozens of these every month, every day. Like the rollout needs to be intense. I'm sure there's a lot that you're learning from from what's going on at SpaceX and how quickly they've made those massive Starship rockets.

2:35:07It always surprises me when there's a new one on the pad because I watched it blow up just a few days ago and they have a new one ready. And it really speaks to their ability to manufacture something at scale, both in terms of speed and just the size. It's one thing when you see the Statue of Liberty. They built that once. But it's a different thing when you see a new Statue of Liberty come off the manufacturing line every single day. So what needs to change at your company in terms of process? What are the requirements to make big things fast and regularly? Yeah. So this comes back to the design of the system.

2:35:42So the system is just this really simple steel shell. It's like entirely hollow inside. Sure. And it's made with mild steel. It's got this axisymmetric shape so that there's a central axis. And the whole system is built around that central axis. So that allows you to use equipment that lets you make these concentric shapes, put them together, weld them really simply. In a lot of ways, I mean, I keep saying simple, but in a lot of ways, it's actually simpler than a car. Oh, interesting. And, you know, this just comes back to the design of it. But what we can go and do because it's that way is we can cost all of the CapEx that you need in a factory for these things.

2:36:18We know how many people will be involved. We know how much land it'll take up. And even now we can say that, look, if we were to go and build a factory of a certain CapEx investment, you know, let's say you built a factory at the scale of a gigafactory in terms of cost, how many gigawatts would that spit out? And, for example, we know that if we built factories with the amount of CapEx that Tesla has put into gigafactories, we'd be spitting out 20 gigawatts a year of capacity. So the manufacturing story is really nice. these are still small enough to like mass produce it's not ship building um and it's not people crawling all over them it's things that you can to a large extent automate so we're actually building our first uh pilot line here in the pacific northwest parts of it will be operational in january and the whole thing should be operational later in the spring and then once that's going then we can really start to to carve and copy that elsewhere yeah it looks and feels so science like it's out of a science fiction book it really it really does it's extremely sci-fi um how do you actually make these shells like i you know tesla has like the giga press or something like that that can form aluminum or stainless steel into a particular shape once you get so big i don't even know is there a bigger machine that just kind of will stamp these things together do they have to be bent or rolled out of some sort of is it what is a mold i don't even know how it gets made yeah yeah it so if you're using mild steel the key thing is to keep the surfaces single curvature as much as possible um so you get to use just like plate rolls and basically they're like big pasta rolls that make you know cylindrical shapes and then conical shapes um and and that's just you know it's just some metal rolls and you push the metal through and it comes out in the right shape then you can put that into your jig or put that into what's called a growing line configuration that lets you align them and then use an automated welding head to weld them.

2:38:15So luckily, a lot of this stuff has been developed. It's pretty well understood. They use it for like wind turbine monopiles. They use it for other axisymmetric shapes like that. So we get to sort of piggyback on a lot of the same stuff. Oh, that makes sense. Yeah, it does kind of have the same shape as like one of those massive wind turbines, the huge windmills that you see. I mean, it's a fantastic project. It's so fascinating to be working in this industry. I'm sure tons of lessons learned and a lot more good stuff in the future. But thank you so much for stopping by. This is fantastic. Yeah, we're going to, our power demands are pretty immense.

2:38:49We're going to need to be doing a lot of reinforcement learning. Expect us to fill out the form on the site to get one of these ASAP. Yeah, awesome, awesome. Well, we'll come back on when we're kicking out those first nodes from this factory here. I can't wait. We love that. Thank you so much. Awesome. Thanks for joining, Garth. Have a great one. Thank you. We have our next guest coming in. Yeah, it's a wild project, right? Just the imagery of seeing people just climbing on top of it. You don't really realize the scale. It's simple. It's simple. It's a small little thing. Really downplaying the size of the thing.

2:39:25Just downplaying it. It's massive. I love it. Anyway, if you're heading to Antarctica or anywhere else to travel the world, go take advantage of some low-cost energy across the globe book a wander find your happy place find your happy place book a wonder even the gold retrievers joined it it's fantastic hotel great amenities dreamy beds top tier cleaning 24 7 concierge service it's a vacation home but better folks use code tbpn when you sign up um do it our next guest is coming in to the studio uh from divergent 3d talking about uh manufacturing plans they have a very cool partnership with talent here and I wanted to talk to Lucas so welcome to the studio how are you doing doing great how are you gentlemen we're great what's going on what do we have uh in front of us so you're looking at the airframe for an autonomous aircraft system so about a six foot long fully autonomous 3D printed structure and if you look in the tail and on inside you see a little bit about divergent which is full functional integration of these airframes to make them faster, better, and cheaper.

2:40:30Wow. Okay. It looks very cool. And the background is obviously green screen, right? That is our factory. That is Torrance, Los Angeles for you. So one of four buildings on this campus. This is where we manufacture, uh, at volume for aerospace and defense and the auto industry. Very, very cool. Awesome. Um, yeah. What, what, who are some of your customers? Any of a partnership with Palantir now, can you explain, uh, kind of the business in, in kind of basic terms and then we'll kind of work through it? Yeah, absolutely. Customers are really OEMs, so really top automakers like Ferrari, McLaren, Bugatti.

2:41:07And then primary focus is aerospace and defense, which are really all the large primes, Lockheed Martin, Raytheon, General Atomics and others, and the smaller up and comers and the larger up and comers like the Andrells of the world and Castilians are also customers of Divergent. And what we have built over the last decade is a ground up platform for engineering and manufacturing. So I want to be very clear, this was a deep tech startup. It required over a billion dollars to build the system. We've got over 750 patents on it. It's not just a quick manufacture and play roll-up strategy. This is really about creating a design differentiation in terms of how quickly you can create performant CAD.

2:41:59Let's call it an airframe, the CAD for an airframe. Can you do that in a week instead of a year and a half? And then how do you manufacture what you've created? And we've created both the source code for that design engine, and then we've created manufacturing process that stitches together our own version of added manufacturing, 3D metal printing, and robotic fixture-less assembly, which allows us to do something extraordinary, which is have a factory which is product agnostic. So you're looking at an installed factory base that can go from making an airframe for a Lockheed Martin to a large unmanned aircraft system for Raytheon to a production vehicle chassis for Ferrari.

2:42:41Back to back to back with no hardware change whatsoever. And that is a new principle. That is a new reality where you are combining cost productivity of mass manufacturing, industrial manufacturing, with the flexibility of what was first thought of as fast prototyping. Really taking those two principles into one system that's the divergent system. We've now scaled to over 18 contracts, really all heading towards programs of record. And to be clear, we're not a prime ourselves. We are a subcontractor to these primes. What we are building is the manufacturing layer and the engineering layer that all of these programs really require to hit the scale that the U.S.

2:43:26needs to compete with our adversaries and to create the deterrence that we need in terms of our inventories and in our capabilities. Yeah, we've talked to a couple of people about this transition from additive manufacturing from a test environment to actual production. Does that mean that traditional subtractive manufacturing doesn't play a role in the parts or airframes that you design? Or do you work with other subcontractors to bring in more traditional workflows to augment what you're actually making so that you're just kind of using the right tool for the job at any moment? Or is there some benefit to being like a pure play in additive manufacturing?

2:44:06Yeah, I'd say that the realities of additive manufacturing costs have changed dramatically. When we started eight years ago, you were not cost productive in the additive space. Today, we're actually cheaper than the caster machine alternative. So yes, you're able to make these very advanced geometries, but you're actually coming in at a dollar per kilogram point that is more affordable for the prime, the taxpayer, the government than the traditional alternative. Where we see the most value is when we can take design authority for the entire, say, end-to-end missile structure, start with the requirements.

2:44:40What is the CFD? What are the load cases? Where are the hard points to the other subsystems? And then engineer that from scratch. That's where we can use high-performance computing, generative design, and essentially optimize across all variables in one cohesive full design package. And that then becomes a 100 % additive structure that is assembled with that robotic system. Are there other systems that end up in these vehicles? Yes, but you're talking electronics, you're talking the payload itself, you're talking about the turbojet or the propulsion source. Typically, when we're looking at structure, we're 100 % additive and we're able to outperform anything that would be cast or machined from a performance standpoint, say 30 % lighter while meeting requirements, but more importantly, in a lot of cases, 30 % cheaper while being able to make$1 ,000 or$10 ,000 per year.

2:45:34To give you some context, one of our 3D printers can manufacture about 400 large-scale munition casings per year. So imagine 100 printers running, you're able to do 40 ,000 munitions per year. And 100 printers running is not an unrealistic installed base. We're running tens of machines already today. And that is what the U.S. government also sees as critical for the U.S. Can we create a civil reserve manufacturing network, a connected set of next generation facilities that can create these munitions at volume? Because to cut it very, very clear, you look at our legacy programs like Tomahawk, JASM, they're not able to hit the rates that are required by our warfighters, by our leaders, by individuals like Admiral Capara, who are closest to the problem.

2:46:26And why are they not able to hit those rates? Partially, because they cannot get the structure. They cannot get the castings at the volumes required, at the quality point, at the geometry that's required. We come in and we solve that in a very dramatic way. Can you talk to me about some of the history and kind of developmental breakthroughs in additive manufacturing. I'm sure there's some stuff that's happened just in the industry broadly, maybe even academic papers or different pieces of the supply chain that have kind of unlocked what you're doing, allowed it to scale. Stuff like what's going on in the alloy space or deeper in the supply chain with parts that go into machines or machines that you're able to buy now.

2:47:08Give me kind of the lay of the land because it feels like we're in a kind of Moore's law moment for this particular technology. You've obviously built a huge business around it. But what else is driving that upstream? Oh, absolutely. Additive manufacturing. It's no secret now that it's going to be supporting the structures in our programs of record. And a decade ago, it was thought of more as an R &D technology. So the industry has moved with us on that. At the same time, Divergent is very differentiated in the cost productivity. We built our own 3D printer. So we use our own machine. we built our own materials.

2:47:42So the aluminum alloys, the refractory alloys that we print with, those are designed by our material scientists. The machine is designed by our engineers, and we actually integrate that machine in-house. So we build it in-house with a US-based supply chain. That machine is 10x the cost productivity of its nearest peer. So even though the industry has moved forward as well, Divergent has led a lot of that movement forward, and now it is printing on a machine that is able to be 3x the rate while being a third of the cost. So a 10x improvement to its near pure. In terms of talking about some of the advancements, when we started, we were printing on single laser systems.

2:48:19So you're talking about powder bed fusion. You've got a powdered metallic on the bottom of the build chamber. You've got a laser overhead. You're going to melt one layer at a time of this part and build it like a layer cake. We're now at a 12 laser system so we've gone from one laser to 12 we're under one kilowatt per laser now we're north of two and a half kilowatts per laser so the amount of energy in this system the size of the build chamber the layer height the speed of our recoding the speed of our gas flow across that chamber our automated build plate loading and unloading has essentially multiplied the rate by more than 20x over the last eight years.

2:48:58You guys developed your own hypercar. What would be the right time for us to start exploring developing a TBPN hypercar given that you just went through the process yourself? You know, if you have a dream and you've got the capital, pick up the phone, call us and we can help with it. That's what I'll say. But Zinger Vehicles, and a more serious note, So really that was about creating a product to create the ultimate tools. And then we created a business out of it. But early on, we said, if we're going to make the best design software, the best materials, the best 3D printer, a real assembly system, you have to create a product and you have to test that product tens, if not hundreds of times.

2:49:40And how else are you going to do that? You're not going to do that with a customer. You're not going to get your iteration. You're not going to get the feedback loop fast enough. We had to print thousands of parts and essentially get it wrong for our first two, three years before we really started to get it right. And then Zinger gave us the systems engineering knowledge, the road data, the branding ability of our own product, taught us about heat exchange, taught us about propulsion. We developed that V8 engine from scratch. We print a lot of that V8 engine. California carb emissions compliant, fully crash certified.

2:50:13That vehicle is tremendous feedback into divergent tools. And now in its own right, it's become a brand. It's the only vertically integrated hyper car OEM in America. And we just delivered our first handful of cars. They're the fastest road legal car in the world now. And there's a reason for that is because it's built on a new technology. It is a different chassis that is in that vehicle than any other vehicle on the road. Did you get any angry calls from some of your customers for just absolutely mogging them? Ferrari's like, hey, we've been doing this for coming up on 100 years. And we're excited to be a customer.

2:50:52But we would appreciate if you didn't just outperform us on every possible level and do it with a fraction of the budget and the talent and the time. Hey, it's motivation, right? You take that car on the track and you know who you're running against. The engineers are there and they see that and then they go to their executives and they say, we got to use Divergent to engineer the chassis and manufacture the parts. We got to do it or we're not going to be competitive. And at the same time, we love competition and so do they. You know, our chief commercial officer, George Biggs, used to be the head of sales and marketing at McLaren.

2:51:22So people are clearly betting on Zinger vehicles as the up and comer in the hypercar space that can truly become the performance car company in the U.S. But to us, competition is a good thing. We're not shy about it. We like to show our customers what can be done. and even if they are, you know, see the competition in it, at the end of the day, they want to make the best product. And if you're giving them the best pricing and the best ability to make that great product, they're going to want to be your customer. You guys might be the most real hard tech company of all time. Not like you being like, oh, we simultaneously are working with like some of the biggest, most important companies in the world from luxury automotive to defense.

2:52:05And then we also on the side for fun and to just prove we made a real hyper car that you can actually drive and take on the track. Yeah, the car itself looks like it's just like a concept car. It's the kind of thing that other manufacturers will put out as a concept car and then fans will see and they'll be like, well, this thing's never going to exist. So I'm not even going to be that excited about it. Clearly a no render respecter. Yeah, no, it's a it's a tremendous design. And the truth of that vehicle is it represents those 750 patents everything about it the way it's packaged the way it's engineered its power density its seating layout its aero philosophy all those differences come from the technology it utilizes uh as well that's amazing amazing thank you so much for coming on we would love to have you back we can go way deeper and uh to be clear we've got a lot of space in the studio if you want to park one of those cars if you need a parking spot here We got, it's gated.

2:53:04We got security. It'll be safe. We're happy to keep it here in the background. Come to sunny Los Angeles and go forward. No, we're here. We're here. Oh, you're here. Yeah, we'll have to come down. Drive it over. All right. We will bring it over and you got to come see the factory as well. Come see the missiles and various structures. That sounds fantastic. Thank you so much for hopping on, Lucas. This was fantastic. Cheers. Talk to you soon. Bye. Really quickly, what do you need if you got a hyper car? You need a watch. go to getbezel.com. Your bezel concierge is available to source any watch on the planet.

2:53:36Their next challenge, make a watch that's so significant. A tourbillon with 75 complications. We'll accept it. I want to see it. Next up, we have Christian from Astronus coming in the studio. Welcome to the stream, Christian. You're technically your second appearance, second appearance because he was in the background. I'm going to hit the gong. I'm going to hit the gong for Christian just because I like you. Hit the gong. Welcome to the show. Look at that. There we go. That's a great game up there. The top down shot of the gong is absolutely, it's just like top notch guys. 1 % better every day.

2:54:14Your camera looks fantastic. The lighting looks fantastic. It's very cinematic. Insane. Courtesy of Jason Carman. Yes, I know. The guy who set up the original Astronis marketing team. Got us all hooked up. Got us the right gear. Set us on the right path. Yeah, it's great. And the green screen, that AI-generated background. Yeah, it's really, really top notch. This is actually going to be YC Demo Day. So they're setting up right now. In our main area, they're going to put YC. Because they're right across the street. So the Demo Day is going to be right back there. They're setting up all day yesterday.

2:54:47We'll see you on Wednesday. We're heading up. We'll be there. Oh, nice. Wednesday. Yeah, so we have a lot to talk about. We'd love to just get the intro and update from you. And then I want to go into where we are on the Relentless Marsh to create the ultimate hard tech vibrial at some point. But first, give me a little update on your side. How are things going? Yeah, things are going really good. It's an exciting time for Astronus. So we build small satellites for high orbits. And it seems like the rest of the world is sort of catching on to this being a thing. We've been trucking away for 10 years at this idea.

2:55:20And over those 10 years, you know, we've built a big commercial business. We launched four satellites at the end of last year. Those are getting close, really close. There's a lot to actually start providing service all over the world. We have five more that are in the factory. But then the big, I mean, the real big update, we have a bunch of government contracts that are really starting to take real flight. So it's everything from just normal communications for the government mission. You know, like same thing out in geostationary orbit dedicated to one spot on Earth. But then also we have this new mission, which is resilient GPS, which is super cool.

2:55:49I think John talked about that last time he was on, just saying, you know, we want to take GPS, break it apart into many smaller satellites to make it more resilient and less able for adversaries to take it down or jam us. Yeah, because there's a ton of GPS denial and jamming that's going on all over the world now, since unfortunately there seems to be a battlefield on every continent at this point. But that's very exciting. Talk to me about startup Vibreels. What's been interesting to you? What's been your favorite? Are we in the post-Vibreel era? Have we peaked? Yeah, we might be, man. Have we peaked?

2:56:20Yeah. I really liked the one from Zipline was cool. I thought that they did like when like the solar punk vibe felt very different. But what do you see and what do you like in these days? Yeah, it's really interesting because you see these companies that have not raised very much money at all, putting like presumably like tens or hundreds of thousands of dollars into a vibe reel to announce them. Like I don't dislike it. I kind of love it. But at the same time, we would never do something like that. Launching a vibe reel early in our job. That's ridiculous. Well, to our credit, to the credit for the first.

2:56:49The first Vibe Reel cost us like$2 ,000. Exactly. Okay, so this is exactly my point. You can tell when you look at some of these Vibe Reels how much money they cost to produce. And you can tell they weren't done in-house. And I think that makes a big difference. Like if you're a company that's doing your own comps, like you guys are just like whatever, renting a car. That was probably your biggest expense. I don't know. Then you make it yourself, like an internal thing, totally different vibe than like we hired a company. We paid hundreds of thousands of dollars because we want to impress you.

2:57:16Like you can just kind of feel that coming off of the screen when it's not made in-house, not organic. And that's more what I object to than the like, you know, just making things look nice. There's nothing wrong with making things look nice as long as you're doing it authentically and you're doing it yourself. So best practices for Vibe Reels for hard tech companies. Give me what is innovative right now? So I think that so what we are shifting to is just more authenticity, more showing, less telling. so we are building we just launched a new astronis manufacturing channel it's pretty sick actually basically instead of it being like you know two dudes yapping away and just like talking about how great their company is which is like the majority of startup comms these days and like you know the founder sits down in the chair it's like the big dramatic opening shot like uh you have that's one end of the spectrum on the other end of the spectrum is an hour-long video of just cutting metal and that's what we're trying to do with our astronis manufacturing channel so we just we've launched like two or three videos uh long form videos and a bunch of shorts that are more vibey and like get people into the channel and it's going really well it's pretty sweet we're getting you know tens of thousands of views on those first videos on a brand new clean channel that's great which is definitely what you want to see getting some good traction totally yeah i do think that we're sort of like beyond the like talking talking head like uh wouldn't like oh wow look the ceo is going to talk like i don't know that's not that exciting anymore yeah yeah yeah yeah i mean there's a lot of companies that will like open source software packages but there's something that's almost like higher leverage about just like hey we have this engineer who really understands how to use this one machine.

2:58:44Let's just put out a video of them talking about best practices for this. It's going to be super niche, but that community is going to love it. And from a talent standpoint. That's very exciting. Jordy, anything? Because I want to know. Well, I will say on this topic, we just had an event here yesterday or Saturday. It's pretty sweet. We had a bunch of manufacturing people from San Francisco all come to the office and meet in person from like a talent perspective yeah like selfishly we had a lot of really amazing machinists and manufacturing engineers and stuff like come check us out but really it was just to like build a community of these folks in san francisco so i mean people often as astronauts they like don't even believe me when i say they were building satellites in san francisco they're like they're like okay but yeah but like where do you build them like like what do you actually like like no it's like literally down the hall from where i'm sitting right now is where we build our satellites but um you know it's cool to get that community together of san francisco people There is manufacturing in San Francisco.

2:59:40It is crazy. Yeah, what are you expecting on the hard tech side from this year's YC batch? I'm sure you've had a little bit of a preview. We're headed over there tomorrow for Demo Day on Wednesday. No, it's going to be sweet. I mean, I think that, Coogan, didn't you tweet this last time? You were like, out is Uber for X and in is like Anderil for dog walking or something. Out is Uber for dog walking, in is Anderil for dog walking. Yeah, unbelievable. Precision guided dog walking. Well, you know what's funny is that we talked to somebody about humanoid robots on the stream earlier this week. And he was like, oh, well, like in manufacturing context, like it's a lot better to just have a purpose-built road rod.

3:00:22Or maybe you want wheels because you're in a factory and you have smooth floors, so you'll just do wheels. And the one example he gave of a really strong use case for humanoids was dog walking. There we go. There we go. I don't know if you want to say it was really strong. What's the dog going to think if it's not a human walking it? I don't know. Yeah, that seems like the one thing you don't want to outsource in your life is like the couple of, you know, the 20 minutes you get to walk your dog and just like enjoy a peaceful stroll. A new idea is a humanoid robot for like patting your child on the back as they're going to sleep.

3:00:49Yeah, I remember last demo day, people were trying to, I mean, this was at a period, I feel like about a year ago, it became popular to kind of dunk on YC. Oh, totally. It's always been people take cheap shots at them all the time, but there were some people critiquing a bunch of YC founders for doing deep tech stuff. Oh, yeah. And it's like, well, shouldn't – is that not, like, amazing, right? Like, do we not want – Also, I mean, that means it's so wrong. YC is sort of damned if they do, damned if they don't. You know what I mean? Like, they're getting criticized for doing just SaaS, and they're getting criticized.

3:01:26Now they move to, like, defense tech and hard tech and whatever. I mean, at some point, they're just doing what they think is best, and that's like the right call. But I mean, as long as we went through YC like however many years, like eight, nine years ago. And at the time there was like basically no hard tech in the batch except like we were the, I think we're actually really one of the first batches. It was us, it was Boom, and it was Relativity Space all in the same batch. Yeah, that's crazy. And so like that was the first real like hard tech YC foothold. And then nowadays, of course, it's like, if you look at their list of requests for startups, it's like 50 to 75 % of the entire list is just hard tech, like defense tech, that sort of thing.

3:02:01uh reactions to wwdc are you going to be multitasking on an ipad do you care at all no i i don't even did they actually release the like clear bubble thing that everyone's complaining about already or is it just like they basically made it they they made they've launched a feature that's effectively like what you could jailbreak your original iphone to do where a lot of the functionality is effectively uh and the ui is like see-through over a background I'm not tracking it. I don't really care. Bullish. Bullish. Yeah, holy man. What are you guys using for expense management these days? What a question.

3:02:39So we actually recently switched to... Ramp. Let's go. Let's go. I love the hat. Wow, he put on the hat. So I'm sitting over finance right now. That's one of my jobs. You are. When I got in there, there was a different expense management system in place, and we had to get the best-in-class ramp. Our dogs are barking. Actually, I didn't even buy the hat. Shout-out to Paki McCormick for sending me a - Paki. There we go, Paki. These ramp hats are hard to get. You're looking at it right here. That's fantastic. Yeah, when we partnered with ramp, we only got one hat. He said you got to work for the second hat.

3:03:16At least you got the jackets. I want one of those. The jacket's coming. We'll bring these up to the bay. We're going to have a bunch of them. We have 412 hats. They're going to go quickly. They're going to go quickly. Not to be precise or anything. It's limited. But you got one with your name on it. Hell yeah. Anyway. Well, this was fantastic. Thanks so much for hopping on. We will talk to you soon. Cheers, Christian. Thanks for coming on. Bye. Good to see you. our next guest is already here in the tbp and ultra drum we have vlad from in in physical am i pronouncing that correctly i'm probably pronouncing it wrong but we'll hear it from him directly welcome to the stream vlad how are you doing what's going on doing great how are you guys how do you pronounce it the name the name the name vlad or in physical or oh it's in physical okay great physical give us the intro what are you guys doing what are you working on it's not the easiest name It's definitely everyone has asked me how to pronounce it.

3:04:13And I've heard so many pronunciations and ways of spelling it. But what we do is open source secrets management platform. So we help developers manage secrets across their infrastructure. And so what secrets are, it's all types of sensitive credentials that basically connect the systems together and get the run software under the hood and no one else realizes that. Very cool. You have to brute. You have to work on. you have to become so successful that people are forced to figure out how to pronounce the name right it's always it's always it's always a challenge and then once you know once you hit that 100 million dollar run rate everybody will will have a good sense of it but um yeah but yeah talk uh give it give us a backstory i know you you announced a 60 million dollar series a from a lot gill on friday very exciting uh not not an easy uh guy to to get on board uh but break down maybe the history of the company and getting to that point.

3:05:08Yeah, no, absolutely. We started in a very open source manner, right? We launched a physical open source project on Reddit, and that's pretty much how it got viral. And that's how it got lots of lots of developers on board. And that's pretty much how it kept running afterwards. And, you know, like at first it was a lot of developers and it started transitioning to a lot of enterprises and very large customers and folks like LG or AI companies like Hugging Face that are using us. We went through the YC batch. So this is kind of like the first stage of the company. This was around two years ago at this point.

3:05:37So, so much has happened ever since. And yeah, since then, I've been focusing a lot on trust, a lot on security, a lot on helping companies, figuring out how to make it simple and straightforward. Yeah. Can you give me a sense of the shape of the problem? Maybe some anonymized or non-anonymized horror stories. How bad can this go when it goes wrong? I mean, it could be terrible. Actually, most of the over 50 % of all the security breaches that happen these days is because of mishandled credentials or some kind of identity issue that was either hard coded or accidentally leaked by one of the developers or, you know, there can be lots and lots and lots of different things.

3:06:20So a lot of the, you know, news articles that you see on different media, whether it's the crunch or whatever, right, a lot of them happen because of secrets management. And the practices that a lot of organizations have, You know, they are trying and a lot of organizations are trying to figure it out. The problem is that the tooling out there is so complex to figure out and the infrastructure of these organizations is also so complex that they can't really do this. And this is how they use it physical. And so the main benefits of this is first security, but also reliability, because it's kind of like hard to realize this, but secrets basically connect all the different services and databases and infrastructure and engineers together.

3:06:59So if you are not able to access secret, even for like a second, it could have lots of, lots of, lots of bad implications because then you're not able to, your users are not able to use your products and software and so on. Yeah, such an interesting place to sit because I can imagine a bunch of knock-on effects from being that deeply integrated into like the dev tools chain. Are you thinking about that stuff already? Do you have an idea of where this goes or what adjacent products might look like? is it is it unreasonable to comp the business to like one password right now in terms of like business model and kind of shape of the product not quite so one password is very much kind of like personal passwords and yeah of course very well and kind of like very kind of like user driven what in physical is we power all the infrastructure basically so everything that's related to software and infrastructure layer so it's indeed a very interesting base because we have to integrate with all the infrastructure tools right one password is very much a lot about kind of like the browser extension, right?

3:07:57Oh yeah, it doesn't need, yeah, they don't need to actually integrate with like Google, for example, but you do. Like we have to integrate with all the cloud providers, like GCP, even like Vercel or Railway or whatever, right? But also all the different version control systems like GitHub and so on, all the CICD systems, basically all the infrastructure components that developers are using. Because otherwise, if we can't integrate even with at least a single tool, then it doesn't provide the full value. So that's where a lot of complexity, what Invisible does grows. And really kind of like what we're doing is we're building out a single platform for enterprises to manage access, to manage secrets, to manage all the sense of information in one place and provision access controls, authentication to all of those services.

3:08:44Talk to me about the open source versus obviously your for-profit company. At least I think so if you're raising$16 million. You never know these days. You don't know. It used to be a given if you went through YC. But then one of the biggest companies of all time became a nonprofit. But what is it like and what is that? Give me the deeper level of that sales motion. Is it just marketing? Is there a clear kind of upgrade path? Who are the best in class? I always go to like Red Hat Linux did really well. But that was kind of a different thing because they were kind of like the company was built after Linux was already created.

3:09:22they didn't create it themselves. Then you have like a stable diffusion as a company and there's a whole bunch of others like Databricks has taken an open source product and then moved it so far forward that they have a differentiated product to offer and they make a ton of money. Talk to me about like the open source business landscape, the pitfalls, like how it works, like what your strategy is. It's super interesting question because a lot of open source business are very different. So a lot of business like Databricks, they basically take existing projects and build things around them. Yes.

3:09:51And physical is in this sense much more similar to books like GitLab, right? Okay. Where we're actually building out and we are the owners of the open source project. And really kind of like what the open source project is about is we want to make sure that any developer and this is very interesting for the for the a lot of open source businesses, right? It's that you kind of like have to target the enterprise components, right? But you also have to make sure that every developer and like a very long tail of developers not even from enterprise, but also students and and indie developers and you know, for weekend projects, they're able to use your tool, right?

3:10:20And they're able to benefit from it. And they are able to do lots of their own tasks with it. And so this is kind of like a fine balance that you have to navigate. And so how physical works is that all of the features that are needed by developers, they're available under MIT license, right? So arguably the most free license out there. And so they can use it for any purposes. They can reuse it. They can fork it. They can do whatever they want. And the features that are actually monetizable, they are features that are primarily needed by large enterprises. right so it's all the different managerial functionality different kind of like a more much more advanced functionality when your team consists of thousands or tens of thousands of developers right and then you you actually have um the problems that you're facing are very very different and and then you can use you can start with a physical open source but a lot of organizations end up transitioning to physical enterprise yeah uh i have to ask because i'm sure people asked it during your series a process but uh is there an ai story here on your web on your website, you mentioned AI, I think, twice.

3:11:21One, with a feature that maybe is AI powered, two, with hugging face, which is obviously a customer. So I just ask because, you know, it's interesting, a lot of my kind of immediate thoughts around what you're building around, like, why build this now? Oftentimes, there's great things to be built that nobody got around to building. But I'm curious to kind of understand the catalysts and how you're maybe benefiting. Well, there's actually a few things, right? So the fact that we are creating this kind of like uniform source of secrets and access and basically all like a single data set, right, for all the access within the enterprises, it helps us create this type of like AI brain behind it, right, where we are able to provision access controls and identify different over-provisioned access and loopholes and everything else that developers are using.

3:12:12So that's kind of like the one part that we're they're focusing and will be focusing even more with AI. The other part is how do all of the agents and how do all of the other AI workers are actually able to operate secrets? Right. Because, you know, we're talking about like they're becoming much faster. They're becoming much smarter in what they're actually doing. But they kind of depend on how fast they're able to access secrets or databases or other resources that they need to be approved. Right. And and right now, a lot of it is provisioned very manually. Right. So this is another area that in physical is really focusing on and how do you actually simplify it and how do you make sure that a lot of these workloads, they can be more autonomous in how they work.

3:12:50Yeah, that makes sense. What's your guys's internal AI software engineering stack? What tools are you getting the most value from today? I mean, in terms of engineering, so people use all kinds of things. I mean, of course, cloud and like basically like all the LLMs out there we are using, it's kind of like a mix of whatever people feel most comfortable with and what they want to choose. That makes sense. Cool. That was great. Thank you so much for popping on. Very exciting. Congrats on the milestone. And I'm sure you already have people breathing down your neck ready to do the B now that a lot is in.

3:13:26So good luck pushing those people off. Thank you. We'll talk to you soon. Cheers. Have a great day. Fantastic. Up next, we have a couple things. We have Alex Heath from The Verge jumping on at 2.30. That's in about 15 minutes. I want to go through some polymarkets about Apple to give us a little bit of an update on what people are expecting Apple to do over the next year. I also want to check in with our intern, Tyler Cosgrove, see how he's doing. The jacket came off. Let me walk. He's rolling up the sleeves. While he's getting ready for his update, I want to run through some of these polymarkets.

3:14:04How much will the iPhone 17 cost? Will I drop one of these? Will you drop one of those? We're down at 10 % now for over$1 ,000. It looks like most people are thinking the next iPhone is going to be$799 or$699. Not a lot of movement there, but it's moving downwards. People are getting more bearish on the idea of$1 ,000 plus iPhone. There's a small polymarket out here about will Apple release a foldable iPhone in 2025? market launched at 12 % it's down at less than 2 % no indication from WWDC that a folding iPhone's coming interesting historical poly market do you want a foldable iPhone no I was I was at the farmer's market on Saturday and was I met another dad and he pulled out his phone and it was a Foldable phone.

3:15:01And he just, he pulled the phone out, unfolded it, did what he needed to do. And I was like, wow, I've never seen one of these in the wild before. But it's kind of cool. Folds up, you know, it's pretty. I'm trying to find why you really need one. I mean, I saw, I think it was Kai Sinat went to China and was looking at a tri-fold phone, the latest and greatest from the top Chinese smartphone company, probably Huawei. Tri-fold. Tri-fold. So it looks just like an iPhone in terms of size, and then it unfolds into something that's basically a tablet. And that is actually pretty interesting. You could imagine that actually catching on in terms of like, you don't even need to bring the laptop with you on the plane because you're just going to fold it out and have a much, much bigger screen.

3:15:49Can you use it when it's in its completely unfolded state? Oh, yeah. You can use it as a phone, and then you can use it as a tablet. So it's essentially both. And the hinge technology is getting to the point where it's reliable, The dust doesn't get in it. It actually folds completely flat. The early phones, they would fold and there would be like a little gap because you couldn't make the hinge fold flat. Imagine unfolding and reading the information. It'd be amazing. It'd be amazing. It'd feel like they had a print edition. Okay. Other polymarkets I want to run through. Will Apple invest in OpenAI in 2024?

3:16:20This is a historical polymarket. It was up at 72 % at one point. Small market, but still it has crashed down to 0%. Obviously beginning to resolve no. But man, maybe they should have been a wild one. The other one that people are tracking these days is, will Apple announce iPhone assembly in the United States before September? We did. We beat them to it. We beat them to it. Let's go to Tyler and see how is the project going, Tyler. Break it down for us. Well, guys, it's done. It's done. The first American iPhone. Manufactured in America. Assembled in America. Wow. You know, it's maybe not the cleanest, maybe not the most perfect.

3:17:00It's kind of a cornflip whether it would pass the inspections. Quality assurance. But you did it. I genuinely can't believe you did it. I was staring looking at the pieces earlier, and I was kind of concerned because I thought we've given Tyler an impossible task. Yep. And we're now three hours and 20 minutes into the project, and it's successful. Yeah, I mean, you know, I'm more of a software guy, machine learning engineer kind of guy. Yep. Not big on hardware, but, you know, this process, I'm incredibly bullish on bringing manufacturing back to the United States. You think it can be done? Okay.

3:17:35If I can do it, you know. Anyone can. Not necessarily anyone. But if you have a college degree. Yeah. We got to kind of get into the economics now and understand, like, you know, Tyler's market value. probably, you know, quarter million a year, taking three hours. We're getting some - You run through it. Potentially some margin problems. Yeah, we've been talking about the$10 ,000 iPhone. I think we might be looking at it. You get an iPhone 4S handmade by a college-educated machine learning engineer. But imagine, you know, in a lot of cars, like the person that made the engine will sort of sign it.

3:18:13Having an iPhone that was sort of signed on the back by Tyler. Why don't we get a Sharpie for Tyler? You can sign the back of that. We can give it away to one of the fans. I'm thinking of starting a boutique, handmade phone company now. I think that's good. Made with love. Just buy the new iPhone, take it apart, put it back together, sign it. Sign it, yep. Assembled here. It's a handmade iPhone. Assembled here. Amazing. Honestly, genuinely incredible work. Really great. They said it couldn't be done. We proved them wrong. And you proved them wrong. Anything is possible in America with enough dynamism.

3:18:43With enough tilers. And we saw it today. With enough tilers. With enough tilers, and we can accomplish anything. Polymarket has it at a 4 % chance that Apple announces iPhone assembly in the United States before September. Not looking good there, but maybe we changed the odds today, folks. And what else is in the Polymarket here? Will Apple release a new product line in 2025? That's sitting at 43%. People are expecting something. Let's go to the comments and see. HomePad plus Home OS is the rumor. Apple Smart Home Display, so something like an iPad that would live in your house and act as the central hub for all your HomePod devices.

3:19:29Very interesting. So, I don't know, we'll be tracking that. Didn't happen today, but we will be getting the full update from our WWDC correspondent, Verge, over at The Verge, Alex Heath coming on the show in just a minute. Well, we can pull up some bangers from the bangers tab. We were Joe Weisenthal. I think Joe Weisenthal had a good one. I don't know if you saw this. There's a startup called Bernie, which claims to have built a reputation marketplace so that you can license out someone else's credentials when fundraising or whatever. They declined to say whether it's ethical. And they say, is this legal?

3:20:07Yes, every engagement on the platform is governed by enforceable licensing contracts. Identity, credentials, and usage terms are verified. Is this moral? So the issue is I'm looking up, I just Googled Bernie reputation. Yeah. And it's just a bunch of stuff on Bernie Sanders. I can't actually find the startup. I was seeing this and I was trying to understand. It's kind of a hilarious concept to basically say like, oh, yeah, you can just use my name to fundraise for your project. Yep. Like that's not going to work. maybe there's some people that would agree to that but sounds social capital now liquid accelerate investor confidence with Bernie's curated marketplace of reputation assets purchase verified team credentials I can't find anything about this company is this ethical that's not for us it says Bernie.market is the website but it could be a prank Bernie.market it let's see it it appears to be a real real website somebody vibe coded it profile characteristics access elite credibility instantly instantly i just don't understand who is this ethical that's not for us to determine credentials are already used to influence funding hiring and trust often informally and without consent we bring transparency consent and pricing to a dynamic that already exists very very interesting so i don't i don't quite understand so this is like you want to build a sas company and you just and you need and you need a harvard educated you need a harvard mba on your deck as like an advisor and so you go i mean i guess like the bull case is this for this is like angel investors and advisors are already kind of like glg exists you could go to alpha sites so advisors i understand and and but this could this could be kind of a like a rebranding of the advisor marketplace.

3:22:02So instead of saying like, hey, we understand the supply chain because we spent$10 ,000 of our pre-seed capital on Tegas talking to experts and we converted a few of those into advisors and gave them some equity, Bernie maybe lets you make that go direct to the advisor relationship. But I don't know. It's very oddly marketed, but it's very funny. Anyway. Extremely sus. Yes. I would, that's it. I would have the founder on just to ask how they expect this to work and not create a wake of problems. Yes, yes, yes. I mean, there are other companies. The founder's like, well, to be honest, I kind of dog - Fooded.

3:22:49Fooded the product and I have no tech experience. I don't have a reputation myself. But, you know, my team has 50 years of experience in reputation. Harvard, MIT, Google, Apple, GPs at Sequoia, Benchmark. Well, I mean, intro.com. Aerospace engineers. Yeah, yeah, yeah. We actually. IMO gold medalists. Many of the OpenAI researchers are on our team. Yep. High frequency traders. We got some Jane Street guys on board. Yeah, Jane Street. We got Citadel to the guys, the team. Johnny Ive. Johnny Ive's involved. He spent one minute replying to it. We're like, oh, okay, okay. So I get it. That makes sense.

3:23:33I'm actually fully bought into this project, and I would trust you with my credentials for you to just sell to the highest bidder. Anyways, we have some more breaking news. Oh, more breaking news. Brad Gerstner's Invest America is being included in the reconciliation bill. Oh. backdated to the beginning of this year. So if you had a child December 1st, 2025 and beyond, they will get$1 ,000. I think it's slated through 2029. That's amazing. So is this like happening, happening, or is it still in the discussion phase? Well, so the reconciliation bill needs to be passed, but it's being included. And they're being called Trump accounts.

3:24:13Wow, okay. That'll be really non-controversial. But congratulations to Brad Gersner. That's very exciting that he got it across the finish line. at least the, he got it across the starting line, I suppose. Yeah. The finish line's upcoming, but it seems like it could, it could go through. I mean, I'm going to be calling these Gerstner accounts forever. Forever. I'm going to tell my, my next child. Yeah. This is, you can thank Gerstner for your college education. Write him a thank you note. Let's pull up this post from Kitsy. Okay. It says, Steve Jobs would have cornered you in a dark alley and beat you with a metal pipe if you even suggested something like this.

3:24:53And this is the new design, this sort of see-through functionality. There's another post from Alex. What's interesting is that X did this for a while and it was not well received, but we ran this experiment. I remember the bottom bar for a while on X shortly after Elon took over. It was like, as you would scroll, it still does it actually. As you scroll, the bottom bar gets transparent and you can still click on it. So it still lets you know that it's there and it kind of fades away so you can see things under it. But I think for a while it was like going up and back and it was kind of jarring.

3:25:33It is funny because... I don't know. I think this could work. Who knows? Some Alex taking the same joke. Or actually he was first. He said, Steve, there's another post showing Steve Jobs, would have taken you out back and shot out your kneecaps for pitching something with this contrast ratio. Another post here seeing how the see-through actually shows up on the home screen. It's funny because you could jailbreak your phone a decade ago. Eric Zed called this out. You could do this exact same thing. I remember it was very cool. It would have some areas that it would be disastrous, but in general, it was pretty cool.

3:26:14Yeah, it happens. Should we go to this news about the doctors? New paper shows a familiar result on LLMs in medicine. Doctors given clinical vignettes produce significantly more accurate diagnoses when using a custom GPT built with obsolete GPT-4 than doctors with Google PubMed but not AI. So we are firmly in the Centaur era. Aaron Levy over at Box sums it up well. He says doctors went from 75 % accuracy without AI to 85 % accuracy when using AI for diagnosis in the future. It will be malpractice for your doctor not to be using AI. This is such a clear societal benefit that it alone should cause everyone to be much more pro AI progress.

3:27:00I completely agree. It's very good. Very exciting. I want to see my doctor go on chat GPT in real time. For sure. Describing what I'm saying. Yeah. I mean, they probably need like a doctor plan where the data is even more sequestered, right? So they're not like leaking HIPAA. Like they need a HIPAA compliant GPT, but that will be pretty easily doable. I guess the question is like there's been a couple AI for doctors companies. And the question is, is that an independent application layer company? or is that something that OpenAI rolls out as a functionality? ChatGPT is the HIPAA-compliant ChatGPT kind of thing.

3:27:38Potentially. No, but there's multiple multi-billion dollar companies focused on the scribe functionality, right? Yeah. Or need. Yeah, yeah. I mean, in legal, there certainly seems to be some decent amounts of traction from Harvey. And so it could be enough of a unique use case that it's not just something that you're going to use ChatGPT for. Cursor for surgeons. Cursors. We'll probably see that. Yeah, yeah, just cut that out, clear that out, grab it, throw it out, put it to the side. Just vibe coding your surgery. You know that like the latest Gemini can just take a webcam view of anything and just be talking to you the whole time about what it's seeing, making recommendations.

3:28:20And you really could do that. So cursor for surgeons. Cursor for vibe surgery. I got vibe surgery. like oh yeah like uh how are you recovering uh it was more of like a vibe surgeon than like uh like a hardcore deep deep research surgeon it was more of a vibe surgeon yeah but it was just a college drop my spleen's feeling it's decent it works for now yeah it might be a bit buggy yeah but vibe surgery i don't know if we're gonna go there but it's possible we're already in the era of vibe legislation vibe engineering vibe coding is very real vibe designing is very real get on figma.com uh speaking of great art uh think word says great art should comfort the disturbed and disturb the comfortable and it's a painting of halo it genuinely is beautiful it's beautiful i want to hang this we should get this uh we should get this painted by hand Tyler your next project you're painting and paint this AI no it's funny because I look at this and I don't think slop I think I think I think cool yeah but again I mean like like what takes this out of out of the slop world is is how it is funny to recontextualize halo this science fiction this video game into this you know uh painterly style of, I don't know, like Renaissance painting.

3:29:47It's beautiful. It's stunning. It's provocative. It's this unexpected mashup. It's the Harry Potter Balenciaga. Two things that normally don't go together. You put them together. That's funny. I would like to see the same. The instantiation of the idea is not what is sloppy, but the idea is uniquely human. And so that's why whoever created this came up with a great prompt, and that's why we're liking it. That's why it's 5K likes. Anyway, Dylan Patel, we mentioned this earlier on the show, but Semi Analysis has a banger post on scaling reinforcement learning. Very, very interesting deep dive. Dylan Patel writes, Reinforcement learning is very inference heavy and shifts infrastructure buildouts heavily.

3:30:24Scaling well-engineered environments is difficult. Reward hacking and non-verifiable rewards are key areas of research. Recursive self-improvement is already playing out. Major shift in 04 and 05 RL training. And so what's interesting is that before in the pre-training era, in the scaling of the big, big transformer runs, the massive runs, Dylan Patel told us this, that the latest Orion run, the GPT 4.5 run, was already maxed out. And they couldn't do a bigger single run without bringing Stargate online. And so the industry has simultaneously shifted to reinforcement learning, which, of course, can be done in a very distributed way.

3:31:01And so you can be generating training data all over the place when inference is low. Like chat GPT use is high during the work week. It's not high during the nights and weekends. And so during those times, they use those GPUs, those data centers, to generate reinforcement learning training data. And the particularly important thing is verifiable rewards. So you have the AI generate a whole bunch of code or write the solution to a complex math problem. But then that math problem is verifiable. You can make sure that the answer is correct. What's hard is when you have non-verifiable rewards. Like, is this joke funny?

3:31:41Very, very difficult. And so there's a lot of reward hacking going on and a lot of work being done on what do you do in a scenario where there's a non-verifiable reward, not like a math problem or some code. And so right now what they're doing is they're having other LLMs grade the results. And so you generate 100 jokes, and then another LLM is looking at them and saying, which joke is funny? And hopefully it learns over time, but it's very tricky. There's also very interesting developments in engineered environments. And so when you're doing reinforcement learning, the classic example is like chess or Go.

3:32:12And so you put all the chess pieces on a virtual board. You play the chess out. but it's very easy to simulate chess because it's just a bunch of pieces on the board. Defined rules. It's a lot harder to create an engineering environment that is simulating code across a database and a web server and a web browser and Python and image generation because there's lag and delay in all of those. And so a lot of the research that OpenAI did on reinforcement learning for video games, they were able to run Super Mario at like a thousand times speed. or they can simulate an entire game of chess in a millisecond.

3:32:49It's much harder to do that when you have to say, okay, go and interact with all these web servers and run into all the captchas and you can't really speed up the environment. And so how you work through those engineering environments that you're running the test on is very, very key. And so there's a lot of research going in there. So I'm sure we'll be digging into this more in the future. Anyway, our guest is joining in just a minute, but let's keep moving on. What should be - You're surrounded by journalists. Hold your position. I can't wait. There's an obituary for Bill Atkinson, one of a kind, the Macintosh pioneer and inventor of HyperCard.

3:33:25He passed away at 74. Very sad to hear that. Very glad that Wired was able to do a beautiful obituary. We'll read from some of this right now. My first meeting with Bill Atkinson was unforgettable. It was November of 1983 and reporting for Rolling Stone. I had gained access to the team building the Macintosh computer scheduling to launch early next year. Everyone kept telling me, wait until you meet Bill and Andy, referring to Bill Atkinson and Andy Hertzfeld, two key writers of the Mac software. Here's what I wrote about the encounter in my book, Insanely Great. I met Bill Atkinson first, a tall fellow with unruly hair, a poncho via mustache, blazing blue eyes.

3:34:06he had the unnerving intensity of Bruce Dern in one of his turns as an unhinged Vietnam vet. Like everyone else in the room, he wore jeans and a t-shirt. Do you want to see a bug, he asked me. He pulled me into his cubicle and pointed at his Macintosh. Filling the screen was an incredibly detailed drawing of an insect. It was beautiful, something you might see on an expensive work station in a research lab, but not on a personal computer. Atkinson laughed at his joke. Then he He got very serious, taking an intense near whisper that gave his words a reverential weight. The barrier between words and pictures is broken, he said.

3:34:44Until now, the word of art has been a sacred club, like fine china. Now it is for daily use. He's so excited about it. Atkinson was right. His contributions to the Macintosh were critical to that breakthrough he'd whispered to me at the Apple office known as Brandly 3 that day. A few years later, he would single-handedly make another giant contribution with a program called HyperCard, which presaged the World Wide Web. Through it all, he retained his energy and joie de vivre and became an inspiration for all who had changed the world through code. On June 5th, 2025, he died after a long illness.

3:35:21He was 74. Atkinson hadn't planned on becoming a pioneer in personal computing as a graduate student. He studied computer science and neurobiology at the University of Washington. But when he When he encountered the Apple II in 1997, he fell in love and went to work for the company that built it a year later. He was employee number 51. In 1979, he was among the small group that Steve Jobs led to the Xerox PARC research lab and was blown away by the graphic computer interface he saw there, the GUI. It became his job to translate that futuristic technology to the consumer working on Apple's Lisa project.

3:35:53In the process, he invented many of the conventions that still persist on today's computers like menu bars. Atkinson, he created the menu bar, Jordy. This is incredible. Which they just today added to the iPad. Really? The menu bar is in the iPad now. He lives on. He lives on. What a full circle moment. Atkinson also created Quickdraw, a groundbreaking technology to efficiently draw objects on a screen. One of those objects was the round rectangle, a box with rounded corners that would become part of everyone's computing experience. atkinson had resisted the idea until jobs made him walk around the block and see all the traffic signs and other objects with rounded corners wow he's like it's it's probably easier for the computer to draw a circle or a perfect square but steve jobs is like no the corners have to be round they will be round and he built it remarkable that's amazing because of atkinson my first experiences with computers were beautiful and magical.

3:36:52Yeah. I was listening to Ben Thompson talk about Steve Jobs and AI and I think he was actually quoting, who was it, Signal or someone on X, one of the accounts that we highlight all the time. And this account had said what would Steve Jobs think about vibe coding? And Ben Thompson's point was like he would have loved it. He would have loved vibe coding and his example was that if you go back to one of the first major keynotes for apple or one of the keynotes i think right before he passed away actually steve jobs introduced garage band and in it he's talking about all the different like uh like instruments that you can use and and he says like now anyone can be a musician and he was really about like just bringing the ability to create to anyone and that's what vibe coding does and it's like the perfect example of that and people get so hung up with uh like the business implications of vibe coding or like what if it hallucinates like well what if yes Steve Jobs wouldn't say vibe code for 20 minutes and push it to production no of course but he would but he would but he would appreciate this sort of just creating software totally totally that process of creation it's not like he's he made garage band being like make the perfect song yeah every time and one of the things that Apple's uniquely behind in, speaking of WWDC, is Xcode.

3:38:15It does not integrate with Cursor. So when you're writing an Apple iOS app and you're writing Objective-C, it's a very difficult language and you can't immediately integrate with Cursor, at least not Apple. But Apple is working with Anthropic. They're starting to work on it. They are solving it now. Anyway, we have our WWDC correspondent, Alex, coming in the studio. Welcome to the stream. How are you doing? What's going on? Guys, I'm at the Steve Jobs Theater. I'm doing this handheld. I hope that's okay. Oh, it's great. It looks fantastic. You're doing a great job. Can you show us what's behind you?

3:38:48Can you give us a little tour? Wow. Wow. The donut. Behind me is the Steve Jobs Theater. Well, it's below me, technically. Yeah, yeah, yeah. This is the dome above it. That's beautiful. Amazing. Give us your reactions. Break it down. What have you seen? What have you liked? What would you put in the bucket of needs still time? in the oven let's see okay I hear I hear you now break it down for us what's going on yeah um I haven't gotten to obviously catch the show yet but I'm curious what you guys think I mean to me the big headline is no new Siri and that means no new Siri for at least a year because that's how Apple works right they're not going to talk about it again in a way that's like expansive for developers until WWDC next year.

3:39:39I think we saw the stocks slide literally as WWDC kicked off because people were hoping for that. I think Apple, having just come out of some meetings that I can't really talk about, but I can just say, I think Apple is wanting people to go, look, we've got all this other stuff. The iPad is more advanced. Be excited about the new design. But I think underneath this is this simmering angst about where Apple is at on AI. And I don't think they have a good answer yet. Do you guys? No, it's, I mean, the internet will destroy them for being like, hey, don't worry about the AI stuff. We made your icons see-through.

3:40:24Yeah, I think there's two things going on. And rightfully so, right? It's sort of, it's somewhat, I wouldn't go so far as to say it's unacceptable because ChatGPT is still great on an iPhone. You can get a lot of value out of it. But I forget who we had earlier. It was like, it would be really great if ChatGPT was integrated with photos and iMessage and all these other things. And so, yes, Apple, from an investor standpoint, I look at Apple, and we've talked about this a lot, Apple missed search, but they get$20 billion a year from it, so they're fine. They can miss AI. Yeah, yeah, sure. and they can miss AI, but it does, you know, we might be able to still value capture is what you're saying.

3:41:09Yeah, I think they can still value capture. But as a consumer, it's easy to be frustrated. And if they auction off this button, the action button to the highest bidder, they that could be a multi billion dollar deal for sure if they act as the front door to AI. Yeah, I mean, I was thinking about the same day as WWDC open AI hitting 10 billion in ARR, right, which is 3x year over year, like on a massive base already of billions a year. And a healthy chunk of that is flowing directly into Apple's pockets, right? Because those are chat GPT subscriptions that people are paying for through iOS. So yeah, you're right.

3:41:42They benefit. I think the question is like, what happens if people start to buy phones less because AI makes other devices more interesting? So like the boogeyman here on the ground today is Johnny Ive, right? It's like, what is he doing with OpenAI? Is Apple scared about it? You can't get an answer either on or off the record. But I think that's the real fear, and that's probably years out. So to be near-term bearish on Apple is probably not correct, but in five years are people buying phones less because we have all these AI wearables that take interactions away from the phone? Maybe. Is Apple going to be winning there?

3:42:18So far it doesn't look like it. So I think that's more like five-ish years out, but that's a serious concern. And the China question is interesting. We had Patrick, the author of Apple in China on earlier today, and he was just saying Apple is going to be fine in the United States in the near term, even if they're kind of missing AI functionality. But in China, if there's phones like Huawei and other players that can offer these deeply integrated AI features, you're going to fall behind in that market. And you have to imagine that the dynamic between DeepSeq and Huawei is going to be radically different than the dynamic between OpenAI and Apple.

3:43:00Totally. That's right. Yeah, that's been a big topic here, too. I was running into another member of the press from China who's here, and he was saying everyone there is super mad that they don't have Apple intelligence, which is like everyone here has it. And it's like, what does that even mean? Wait, so they didn't even roll it out? No. They didn't roll it out? No, it's blocked. It's blocked by the regulators, and now it's caught up in the trade war there. So the regulator I was just reading the Financial Times actually had a great report on this how the local regulator was blocking it because of the trade war and So I was caught in all these geopolitical tensions that are really messing up their AI strategy and to not be able to sell the marquee feature of the phone You shipped last year in your second largest market.

3:43:39I don't know. So maybe China's their first largest I don't know by users. It's probably close. That doesn't seem good. And so that's that's a big question. I have yeah Yeah, yeah. I think the other interesting thing is that, like, this is a developers conference, but all the consumers are watching and consumers are the ones that are, I feel like they're the noisiest online about various redesigns. I'm more interested to dig in as we get more details this week into some of the olive branches that Apple could extend to the developer community on the AI side. Because even if they're not going to give up the Siri button to ChatGPT, there are a bunch of different small LLMs and different summarization functionality and different access to parts of the chip that could be used at a lower level by all companies, including upstarts.

3:44:28They just want to build a cool app that maybe needs a little more horsepower on the AI side. And the chips are great. And so giving people more access to that at a deeper level, that feels like a narrative that could be completely drowned out by, oh, we didn't get new Siri. But if the developers are increasingly happy and see that there's more opportunity to just build cool things, we might just wake up in a couple months and be like, oh, wow, there's this amazing new app that's clearly using the chips more effectively. Like, my phone feels faster. It feels more advanced. And it's not an Apple product, but it's enabled by the way Apple kind of, you know, opened up the ecosystem just a little bit.

3:45:04Even if they didn't go full, anyone can build the next Apple. Yeah, I mean, that's right. I think opening up their models that they announced today to developers could be a big deal. But it remains to be seen if the models are good or not. I had a developer come up to me who's building this like AI social network. And he's doing everything on device because he wants it to be privacy centric. And Apple's perfect for that. That's what they do. It's like their models run locally. they're probably not the most performant but they're gonna they're free and they run locally so for him it's a win he's an indie developer trying to build a cool ai thing is that something that huge apps are going to use i doubt it yep so is it going to be enough to shift the narrative on like yeah apple's good at ai now for developers yeah i'm not sure i'm kind of skeptical what's up with that uh research paper that an apple intern put out it was an intern i think he was an intern did you yeah any chatter about that they basically said uh rl is fake yeah shitting on reasoning yeah uh well it's like shitting on the test time compute stuff so i was basically saying like which it kind of makes sense like these are these are fancy parrots at the end of the day right they just repeat things back so if you just like layer on a bunch more of that does it actually get to some semblance of intelligence like i'm totally down to buy the most fancy parrot though give me a fancy parrot i would buy a fancy parrot today i think that's what apple's discounting is like how much we love fancy parents totally yeah i don't care if it's actually thinking or actually sentient like did it book my flight or not did it turn on my alarm yeah it's the argument of like time the android user being like the iphone is not the best phone it's it's the camera on the android is you know a little bit better it's like do the pictures look good or not yeah when i take I'm trying to make sure my dog does it.

3:46:44Is the exposure like roughly correct? Yeah. Is liquid glass hot or is it ass? How do we feel? Is it liquid ass? John doesn't get the. Yeah. I mean, it seems like, I don't know, like people were kind of like ready to chomp at anything and being very like negative. I don't think it's as unprecedented as anyone has ever said because like this has been tried in other apps. It does kind of work. I'll need to kind of play around with it. Do you have a chance to play with it? I think it looks, personally, I think it looks cool. I remember jailbreaking early iPhones, iPod touches to have that functionality and it was cool.

3:47:24I never once in the last 10 years have been thinking, I wish I could go back to that jailbroken UI that I had when I was a kid. So I don't give them any credit. Like this is the kind of thing that they should announce at WWDC and move on quickly and not try to be like, hey everybody look at this. She revolutionized it. I would just be very surprised if Apple whiffed on like a UI design feature because that's like what they're great at. I'm not – like the unexpected upside is like can they train a huge foundation model and build a gigascale, hyperscale data center? Like that's kind of out of their wheelhouse new territory.

3:47:59But I would imagine that whatever they do on the UI, UX side, like it might be a little jarring. Like when they went from skeuomorphic to flat design, there were a lot of people, oh, I like the old one. But then you get used to it and it's fine. So I would imagine that in the fullness of time, we all get used to this and it's no big deal. Although, you know, people are still upset about the Photos app. I feel like I'm getting used to it now. It's still like pretty rough. Some people aren't updating, but it hasn't been as jarring. I've gotten used to it. I don't use any of the new features yet.

3:48:28Yeah, it feels like I've retrained myself on a bad, what I think is a bad user experience. Yeah, it feels very search built with the Photos app. It feels like it's designed for you to go and just start searching. but the actual categorization of images and videos isn't quite there yet so if you're looking for like oh yeah like there was a video of my dog but there was a newspaper in the video i should be able to just type in like video of dog and newspaper and the actual ai that's running to tag that stuff isn't quite there is not there and people think of your photos are for capturing memories and people think of memories in sort of a yeah yeah an order yep and they're going to kind of go back to the moment in time where they capture that moment.

3:49:10So I don't think they've nailed that. Yeah, yeah. Anyway. Yeah, the only other thing that stood out to me was Vision Pro. They're really positioning it as like an enterprise device now. Interesting. And I think that's because that's who's willing to pay that price point for it right now, right? And so it's like, you know, some of that stuff's cool, but I didn't sense a lot of energy around the Vision Pro beyond that, right? Interesting. And I think that product is still kind of a question mark in terms of like, is it going to be a bigger thing or not? And is that positioned as like, hey, we just want you to buy these for your employees so they have a bunch of monitors?

3:49:49Or like, yeah, like you're working on an assembly line or something. That's so odd. Oh, interesting. That's so odd because apparently they pulled the plug on a enterprise-focused device. Tyler, would you have benefited from having an Apple Vision today? Tyler built the first American-made iPhone, our intern. We had our intern build an iPhone. He assembled it in under three hours. I don't know if you can see this. Would Vision Pro have helped you today? Graphical overlays while you're building? I think it would have helped. I mean, I didn't have any instructions at all. So I think kind of the bar is just any instructions.

3:50:20Putting the instructions overhead. Vision Pro, you know, I think it would have been helpful. It could have been good? Okay. Long apple. Not investment advice. But my takeaway on Enterprise Vision Pro is that it's odd because I heard that they were building a version that was specifically for pairing with a desktop setup and multi-monitors. And that was kind of the one direction they were going. And Ben Thompson was writing, like, that's what I want. That's the perfect product. I've always thought that the win, like, they have to win in one category before they win in all of them. Like the iPhone first and foremost was a phone and it replaced a phone and then it earned the right to win everything.

3:51:03And the Apple watch first and foremost was a watch and then it became a fitness tracker and everything else. And so I've always thought that the Apple vision pro should have been, hey, we're Apple. We've never built a TV. This is a replacement for your TV. It's going to be really, really good at Apple TV, movies, IMAX, 3D. Like this is what you buy it for. It comes pre-installed with a ton of movies and TV shows. We're not even shooting 3D movies or spatial video. That's not even relevant. It's just a device, and we're just going to super optimize it for watching movies. Because I got one. That was my favorite thing to do on it.

3:51:39But it wouldn't work in the dark. So I would turn off the lights when I'd go to bed, and my wife would be sitting next to me. And I'd be turning the headset on and off because it would lose the tracking. and it would pop up like your vision pro is not working right now. And it would clearly wasn't, it'd never been used in just like day to day when people watch movies. And so I feel like it still doesn't work at night. It doesn't work in a pitch black room because it has cameras that track the world around you. And so, so you can't even watch a movie. Yeah. Yeah. And then, and then the plane mode was a little iffy too.

3:52:12And they had a special, they had a special like mode that you could flip it into, but it just, it was trying to do too much. And I feel like they should have just been like, this is a home theater for people that can't afford a home theater. And that's a lot of people. And that's a lot of iPhone owners. A lot of iPhone owners can afford a$2 ,000 TV, but they can't afford a whole cinema room in their house because real estate's expensive. And I even heard that one of the leads on the Apple Vision Pro project was from Dolby. It was a Dolby executive. Yeah, Mike Rockwell, the head of it is ex-Dolby.

3:52:42That's right. And so I thought it was like, okay, just lean into that fully. Think about what the The experience is you put this on for two hours, you watch a movie, but there's a whole bunch of different decisions that you make if that's the case. And they clearly kind of did 90 % of that and then 10 % of the other stuff. And now they're all over the place. And, uh, and, and I still just want the home theater on my face. Uh, did you guys see they showed two people wearing the vision pro in the same room on the same couch, watching a movie together? I didn't see that, but that was like something they showed in the keynote.

3:53:12Everyone hates that, but I actually think that's totally reasonable. I don't think that that's crazy. I think it's fine. I think it's fine. We're all just alone together? Yeah, yeah. No, no, no, no. Because you look over, you see the person. It's the same experience as watching a TV on the wall. It's the same thing, but it's just a bigger TV. Okay. It's great. You can still, as long as you can hold hands. I'm super bullish on that. Okay. Yeah. This is a weird take. The bull case for that is that somebody can get up to go to the bathroom or get food, and they don't have to pause the movie. Yeah.

3:53:42No, no. I think it's the future, but they need to make it lighter. They need to make it cooler, like maybe some sort of fan because your face gets hot. Like there's a lot of there's a lot of industrial design challenges to get it to be like fun to actually watch a two hour movie or three hour movie. It's still in the demo phase. You put it on for 15 minutes and then you're done. Are you going to be moving over to the iPad for everything? For all my content creation? Yes. I thought that was actually really interesting that they position the iPad as like the podcaster device instead of the back.

3:54:12Shots fired. Shots fired. Maybe for weekly podcasts. I don't think we could run that. Man, I like my iPad. I just, until they invent a way on the iPad to multitask as well as the Mac, where I can tab between things fast. Well, so that was a big focus. I haven't seen it entirely either, but the big focus was on multitasking, allowing you to have multiple windows open at once. But I don't know if still. Who knows? anyway thank you so much for hopping on this was fantastic uh go get some scoops go get some scoops that's what i'm here for amazing bye to the spaceship scoop lord it's beautiful bye spaceship honestly the grass is looking amazing yeah there's that nice out cheers alex thanks for coming on good to see you later bye um i am deeply disappointed that they did not launch a mahogany iphone that would have been the best the official would have business the official would wood.

3:55:13Yep. Um, liquid glass. I'm watching the video from Tim cook here. It looks good. I don't know. I think it looks cool. I think it's going to grow on me. I think it's fine. I think it looks cool. Yeah. I just don't think, and I'm, I'm excited to play around with it. I just, Apple's going to whiff on that is so ridiculous to me. Like that, this is, this is all the best people for this. They whiffed on photos a little bit. I think you'll get used to it over time. Anyway. Um, this is fantastic. Thanks for watching. Uh, leave us five stars on Apple. Wait, we got to talk about some reviews. What else do we got to talk about?

3:55:45We got some reviews. Oh, we got some reviews. Let's read some reviews. I'm going to be pissed if there's no ads in here. Please. Silicon Valley's Morning Espresso. If NPR did CrossFit and binged Red Bull, you'd get TBPN. Let's go. It's the daily show for technocrats hosted by John Coogan and Jordy Hayes, two gentlemen so classy and caffeinated I had to double check they weren't simulations. Every weekday, TBPN slings the latest hot takes, venture insights, and latest in AGI land and crypto faster than you can pop a Lucy. one minute it's fortnight's in-app economics the next it's an ai belt and road geopolitical manifesto and somehow it all works it's like if all in had a baby with the daily show and then fed that baby a steady diet of easy term sheets and synthetic dopamine 10 out of 10 would recommend even if this review is written by an llm trained exclusively on elon tweets and founders fund pitch decks tune in before your gpus get repossessed that's fantastic great um new favorite podcast by mile by Flynn amazing show amazing host amazing guests can't get enough and keep wanting new episodes to drop earlier even though it drops every day for three hours we're working on it we might we might change the time we're on we're in hour four now so hooked on tbn by uh tbpn by harry campbell i've been hooked on tbpn lately and really loved the episode that featured the waymo toyota partnership that was when we were traveling i said you were like yeah they have a partnership with uh toyota and i was like oh you get like a supra that whole bit about like the waymo supra yes it's like the tech world's version of sports talk radio in the best way possible john and jordy keep things fast-paced unfiltered and genuinely fun but still manage to dig into serious stuff happening across startups ai and venture capital you can tell they've actually built stuff themselves which makes their takes way more insightful than the usual service level chatter whether it's breaking down the latest funding around or grilling a big name founder it always feels like a real conversation not a polished pr piece if you're into tech and want to stay sharp without falling asleep this shows it thank you harry kind of kind words absolute legends of technology and finance i think um from d bearson bass takes gong hits and vibes of techno capital abundance are a few words i'd use to describe the show really enjoy it seriously and they do a great job curating guests some of which are now in my ex Rolodex.

3:58:04It's time to accelerate. Love it. Anyways, thank you for all the kind words. Folks, we love each and every one of you and go leave us a review, but please put an ad in the review for your business or your company or your fund. Put an ad in the review and we'll read it out here on the show. Everybody, have a fantastic rest of your day. I'm going to go hit this gong. Take us out with a gong hit and then we'll check it on Tyler. Tyler. It's a historic day. Tyler Cosgrove, intern at TVPN, has assembled the first iPhone in America. He did it, folks. He did it. He did it. There are no pieces left with scraps.

3:58:41In a cave with scraps. He did it in a cave with scraps. Is it signed now? Show us off. Show us what you got. Oh, he signed it. There we go. Right there. Yeah. Let us know if you want that. We'll mail it to you. Tyler. If you've been listening for four hours, you deserve it. We knew you were cracked. We didn't know you were this cracked. This is fantastic. Thanks so much. Absolute legend. We will see you tomorrow. We have another great lineup. Have a great day. Enjoy. Have a great afternoon and evening. Cheers.