Applied Intuition: The $15B Physical AI Company (You Should Know)

[00:01] Mark and Dreesen has been with us from day one. Series B was led by Hemant at General Catalyst. Series C was Bilal at Lux and D was Elad Gil. Without a doubt, our customers look at the fact that we've been around for nearly 10 years, our crazy claim to fame that we've never spent the money we've ever raised. We have to talk about that. Most companies in the Bay Area are more hobby projects than they are earnest, serious businesses. [00:30] here. People don't want to work in mines in far away places. These are dangerous jobs. AI really, in these industries, is being pulled out of our hands because technology can help solve this problem. We opened also in India and UK offices last year. About 18 sites globally. Australia could be one of those countries where we're doing driverless trucks and we're doing a bunch of different ports. We're kind of hopscotching like a board game throughout the planet. At 10 days, we showed them a video of these infantry squad vehicles driving autonomously out in a test [01:00] take a long, long, long time and it requires many years of planning. It's just false. [01:15] Molly, what's the craziest thing I can do to get this thing to be viral? I don't have a sword. Kassar, what is your hottest take? [01:24] My hottest take? I've gotten in trouble for giving hot takes, but you know what? I'll try to be restrained, but maybe I won't. I think most...

[01:35] companies in the Bay Area are more hobby projects than they are like earnest, serious businesses. And it's kind of like, we'll take a swing on this. And I don't mean to disrespect my own co... I'm in the fraternity of co-founders or founders who are kind of toiling away. And it's very hard to see who's doing it for the right reasons, doing it for the wrong reasons. And I think it's kind of maybe some of the wrong motivations that exist. I think if you roll back the clock 15 years ago or [02:05] were entering the Bay Area, literally around that time. I'm not like one of these people where the old days were always great and the new days are terrible. [02:12] but [02:13] the new days are definitely popular like it is like i think there's some time between like the social network and the hbo show silicon valley it's and then like just us as consumers just adopting smartphones and software becoming a real part of our day-to-day life silicon valley like sunnyvale went from this cottage industry to suddenly the center of you know economic activity [02:43] magnet. And so the valley today, especially, you know, I always make the distinction between South Bay and SF Silicon Valley and San Francisco, but you know, a lot of times it's interchange because I think the cultures are different. [02:55] Yeah, I think the cultures in Sunnyvale and Santa Clara and Mountain View are different than the culture in Mission or Dogpatch. I do. Maybe it's a perception and maybe it's not true.

[03:08] But I don't know, at least I perceive it. And so the point being or the punchline being is that the [03:15] If you look back at [03:17] um finance in the 80s and 90s you know manhattan attracted everybody and it kind of became a money thing and [03:28] I'm not to say that the old Silicon Valley wasn't a money thing. These are businesses to make profits. [03:34] but it seems like it's more of a money thing uh and is that good or bad i mean who am i to judge but [03:42] At least that's not our motivation. I think that's singularly your motivation. You know, there's the [03:47] I think you probably maybe build a company a different way. Like we wouldn't build a company this way if just making money was the only outcome we wanted. So I actually wanted to talk to you about that on like a couple of different dimensions. One of them being, I feel like you just came out of stealth. Yeah, yeah. Well, you know, cult classic, not bestseller. [04:09] So I think if you know, you know. Mark Andreessen has been with us from day one. [04:17] person. So I think, yeah, it's his guy. He created this thing called the web browser, which is, yeah, yeah. But he's been with us from the beginning. And, you know, our Series B was led by Himant at General Catalyst. And, you know, it's like where Series C was Bilal at Lux and, you know, D was Elad Gil. These are like this, you know, they're the in folks of Silicon Valley. And I worked at Y Combinator before. I worked at Google before. So we were here. I think we're just, our relationship

[04:47] different. It's not been antagonistic or anything like that, but I think... [04:51] We have actively sought that limelight, I think, before. That's changed in the last year. Because now every time people hear me say that, they're like, oh, but you were on this podcast. Yes, we're actively changing that. We're here. We're talking to you. That's something that we wouldn't have done a couple years ago. And we had success a couple of years ago. I think there's two practical reasons why we've changed this. One is just recruiting, recruiting numbers. Historically, we've gotten all the people that we've ever wanted. [05:21] you have certain like benchmarks as the company grows. And as we've gotten bigger, the benchmark moved to where we're doing, we're still doing a disproportionate amount of outbound to the stage we're in. As you roughly as you get bigger as a startup, your outbound actually drops because people know who you are and you're starting getting a lot of inbound. A company like OpenAI or Anthropic, there's other reasons because we're also a technology provider and they're more of a consumer prosumer product. They're not going to do a lot of outbound. There's enough inbound coming [05:51] And we're, you know, all of our values can be reduced to two words, radical pragmatism. And so it's like, [05:57] that's maybe not the right strategy. And then the second thing is that I think, you know, heavily influenced by Mark and some of our investors who said, let's get the word out there. I think this is a great company. And I think earlier on also, you know, we were very, very thoughtful of not creating so much attention that would inspire a lot of people to compete with us. But, you know, [06:17] Now it's kind of like we've grown up a little bit. We got some muscles. You know, we've gotten a few few rounds under the belt and we can go into the into the ring and fight basically any any any oncomer. When you're really small, you're embryonic. That's a little dangerous. And I do think actually the new crop of Silicon Valley companies, some do take inspiration from us in the sense of, you know, not like the pithy things like the shoes and the name and all of that stuff.

[06:47] Maybe it's just perception, but it harkens back to that more of that craftsmanship aspect of building technology and building businesses and building companies. It also anchored back to our company's mission, right? So our mission is to build physical AI for a safer, more prosperous world. And you sort of hit a point where... [07:06] to accomplish that mission, we need to be better known. We need to have access to a wider talent pool. And so that's why we've been more open. Yeah. And honestly, our customers do know us. I mean, we are very well known in our industry. I think we're more well known in construction and mining in Asia than maybe some bar in the Indian Mission. But that directly correlates to where we're going to make money, right? Pretty cool. Yeah. Yeah. So, Peter, you were on stage and [07:36] and some of the biggest milestones. For people who didn't get to see that, maybe they will, but [07:41] Could you share some of those biggest milestones? Yeah, so we started in 2017 in, I call it the early era of self-driving cars, and back then, [07:54] For strategy reasons, we didn't think it was going to be smart for us to really go directly into self-driving at that moment. But we started out building tools. We built simulators and data management offerings and large-scale distributed compute offerings. [08:10] and we sold these out to the industry. That was our early success as a company. They gave us resources. Yeah, and we just learned and learned and learned. We really expanded out and became very horizontal. We started working in a bunch of different industries.

[08:27] And we were very much on top of all of the latest AI breakthroughs at every step. [08:33] And so we could sort of see when something interesting was happening, should that actually change our strategy? And so there were a few interesting things that happened which evolved the strategy. One was transformers that were originally successful for large language models, like you heard about Anthropic and OpenAI. Those started to have an impact on self-driving technology and robotics. [08:56] And we thought that was really interesting. And then on top of that, there was some breakthroughs in what's called end-to-end deep learning, which is where you can now take lots of data and you can train these models to actually control these systems in a way that wasn't possible before. [09:26] any kind of machine and that's what we were showing downstairs. That's a bit of the history. [09:31] Sorcery is brought to you by Brex, the financial stack trusted by more than 30,000 companies, including one in three venture-backed startups in the U.S. Nearly 40% of startups fail because they run out of cash. Brex is literally built to help founders avoid that. Unlike traditional banks that let your money sit idle, chipping away at it with fees, Brex is designed to help you spend smarter and move faster.

[10:01] powerful account. You can send and receive money globally at lightning speeds, get 20 times the standard FDIC coverage through their partner banks, and even high yield from day one. With same day and even same hour liquidity, access your funds anytime. Companies like Scale AI, DoorDash, Service Titan, HIMSS, Anthropic, Flexport, Robinhood, and Plaid trust and use Brex. [10:31] Turing is training the next generation of AI with tasks that require real expertise and real world judgment. That's why companies like NVIDIA, Anthropic, Salesforce, and Gemini partner with Turing. Turing builds realistic reinforcement learning environments and data systems based on real operational traces. The kind of infrastructure frontier labs need to train superintelligence. Visit Turing.com slash S-O-U-R-C-E-R-Y. [11:01] When are you going to the moon? [11:02] Oh, you know, it's funny. We actually did work on moon simulators. Yeah, we did. [11:08] We did. So we go there in a simulated reality. Yeah, I can't remember exactly the customer at the time, but there was somebody that was developing a moon rover and they used our... I remember them, but I don't think we're allowed to say who. Yeah, it was like our vehicle dynamics... [11:27] simulations for simulating like literally the moon rover driving on uh on the moon surface so we have yeah wow oh my gosh so when you work with a new customer or a new industry or vertical how do you start to get integrated and onboarded within them uh it really depends i mean we're in a bunch of the verticals so maybe it's better to look back and look forward because we're we're in a bunch of them uh already

[11:52] It's a mix, and this is more for like education for founders who are trying to do this. I'll put it within that context. What's best is first if you know some vertical that establishes a pattern. So I'm, you know, both of us collectively know the automotive industry. [12:05] extremely deeply, right? So then you have a pattern. And then you have to, defense being the next one that we went into. We saw some inclinations of like, okay, well, we can sell to like a General Dynamics as an example. And they'll use the tools kind of like General Motors. Once you have your foothold in there, then you start very quickly understanding, okay, these are the types of people with these are the types of backgrounds that are going to be successful in this environment. And then we hired our, which became our gov team. First with just one person, and then ultimately [12:35] I think founders sometimes make this mistake where they want the perfect [12:40] lead for the Tokyo office. They want the perfect lead for their, you know, whatever, construction or mining business. And perfect is often the enemy of done. And so it's, the ideal is [12:51] the first person ends up becoming the lead and is the lead for the rest of the years. But it's better just to start. And so we started, I mean, all the things that you're seeing now, we're seeing these are compounding results for many, many years. [13:04] There's no fast way. One of the things that I think is underreported in physical AI is how this technology will kind of diffuse to the real world. Whether you're working in kind of anything from consumer AI to code complete products, all that diffusion happens, all that distribution happens through browsers and phones. And everyone has one of them, and they're all fairly consistent environments. Physical AI is very different.

[13:34] The regulatory environments are different, the machines are different, the cultures are different. And so the diffusion is actually way, way slower. [13:44] Time becomes a much more important thing. Without a doubt, our customers look at the fact that we've been around for nearly 10 years, our crazy claim to fame that we've never spent the money we've ever raised. We have to talk about that. I was just thinking about that. So you've raised over a billion dollars. First of that, yeah, in primary, yeah. Because we've also been tenders in secondary, yeah. And you're saying you haven't touched any of that money. [14:14] the truth and we're not using our our capital to like pay payroll and things like that um yeah which is kind of crazy but for a customer [14:21] who wants to have this long-term relationship [14:24] That's really confidence inspiring. [14:26] They're like, hey, this company has been, they got a thousand engineers, been in business nearly 10 years, and they have a functioning business. Like, we can do a five-year relationship with them. We can do a 10-year relationship with them. And that makes it, you know, frankly, difficult to compete with us because, you know, we've been in the business. But like I said, that's very relevant. It's not relevant if you're building, you know... [14:47] Anthropic, you're building some open claw kind of product. Because that doesn't matter. [14:51] But in physical AI, where the diffusion is way more complex and multivariate or variegated, I should say, then suddenly time becomes an important thing. Staying power becomes important. So what is it like, we talked about this a little bit, but what is it like hiring out in this AI era versus before? And what are you looking for in those hires? The funny thing is, I mean, the people who come to Applied are frankly different than people who go to foundation lab companies.

[15:21] I don't know if you saw our research talk. We are as technical as any AI company in the business. And a lot of people, I think, because they see our commercial success and they somehow think we're a really commercially oriented company. This is as technical of a company as you can imagine, all the way to the big labs. Our commercial teams are actually really small. Our commercial teams are surprisingly small. So then why would an engineer choose us versus, [15:51] working at like a cool company in San Francisco, and I kind of am implying we're not cool. Really, what I'm implying is we tend not to attract those types of people, particularly because the type of technology you need to know and be interested in is different. If you like sports cars and you like fighter jets, [16:11] Applied is a great place because those companies aren't doing sports cars and fighter jets. Or if agriculture resonates to you because that's your family's upbringing or construction or mining, those are the people that fit that Venn diagram, that overlap of they love AI and they want to make the farmer's life in America easier. [16:32] that's a great candidate for us. And so I think in that way we've done quite well. And also just like, you know, what I was saying earlier about us being more public, [16:40] It's not necessarily that we've had... Everyone has a talent, you know, everyone's fighting for talent. The real point is, I think, the more we get our message out and how distinct the company is, the more it resonates to those specialists.

[16:53] who are maybe working at Google right down the street from us and don't even know that we exist. They're like, I love the car business. I grew up working at Ford, and then I came to the Bay Area and worked at Google for the last 10 years. That's a great candidate for us. That candidate is not as relevant for, like, Harvey. [17:09] And we could both really, all of us could really love AI. And you have plenty of runway. Yes, exactly. But yeah, at a high level though, there's honestly two types of people that we're really looking for. One is it's the enthusiastic engineer who [17:27] loves AI tools and they can use them super effectively. I think it's becoming a little bit less important now to be, let's say, a specialist software engineer in a given field because the AI tools, they can turn you into a specialist much more quickly. But you need to be an expert in the AI tool use itself. And that is a real skill. And so we test for this hard now. And we're looking at people who are just really, really good at using AI tools. [17:53] And then the second type of person relating more to the research is individuals who so deeply understand how the AI systems work that they can actually push forward the state of the art. And that's really where you get into the deeper research topics. But it's that combination of deep research plus just experts in the application of these tools. And that's what makes us a strong engineering organization. [18:17] Is it true there's over 40 CTOs? [18:20] Yeah, we have ex-CTOs or co-founders of that company. It's a very entrepreneurial culture. Yeah, which sounds...

[18:28] so like flat and dry, but it's absolutely true. We attract a lot of founders. I think, like I said, cult classic, not bestseller. I think among founders, and I'm really proud of this because I want us to be, [18:43] founders founders you know i i think to some degree i i like the respect of other co other founders um [18:51] is that I think my YC background and us building this very intricate business and all these different verticals globally, I think the founders know how difficult that is. And they know that that's a special thing we've accomplished. So I think the magnet has been like, [19:07] this company has somehow figured out how to take really complex technology and, you know, have it go out into the world and all these different places. And maybe I can learn from that. And so if you're a [19:17] CTO whose company isn't working. [19:22] We're always hiring. Applied.co slash careers. Or you can just email me directly. Oh my gosh. [19:27] So today you launched the Applied Edge product. I think, again, I think your defense, well, I think mostly all of Applied Intuition is a bit underrated. But I think in the defense space, at least for the public, it is. So there's a stat that you launched an autonomous infantry unit in less than 10 days. Is that correct? [19:52] Yeah, that's right. Yes. So Secretary of the Army, Dan Driscoll, visited our headquarters in this campus about a year ago. And- Oh, it was the old office. Oh, you're right. Oh, yeah.

[20:05] And he... sort of a throwaway comment. He sort of said, "Hey, what if I gave you an ISV and a Humvee? Could you guys do something with it?" And I said, "Yeah, I think in 10 days we could do something pretty interesting." And to be honest, I wasn't expecting much, but then after he left, we got a phone call and it was somebody in the army saying, "I need an address and delivery instructions." [20:35] And so, yeah, in 10 days, we had a very small team retrofit our vehicle operating system and our SDS, or self-driving system, onto the vehicles. And then, yeah, at 10 days, we showed them a video of these infantry squad vehicles driving autonomously out in a test environment. And then they shipped it off to an Army test facility, and soldiers got to use it for a few weeks. So it was just a great example of adapting commercial technology into defense. [21:05] has spoken about this event actually on many occasions now, because it's a pretty interesting story. [21:09] Yeah, there's a lesson there, which is, I think this view that everything is going to take a long, long, long time and it requires many, many years of planning is just false. And I think especially within the reality of today, [21:25] This technology needs to get out there faster and faster and faster. We don't have the luxury of waiting three or five or 10 years. VCX by Fundrise, the public ticker for private tech, allowing investors of all sizes to invest in venture capital. View the portfolio at getvcx.com. That's getvcx.com. Some of you may not have heard this yet, but our sponsor Public just launched something called Generated Assets, and it brings AI into investing in a way I've honestly never

[21:55] You type in an idea like AI-powered supply chain companies with positive free cash flow, or defense tech companies growing revenue over 25% year over year. Publix AI then dispatches a swarm of agents that scan every single US stock, evaluates them, and instantly builds a custom index around your thesis. What really stands out is how clearly it explains why each stock is included. And before you invest, you can even backtest your idea against the S&P 500, so you're making decisions with real context, not just guessing. [22:25] For more curated assets, Public lets you invest in stocks, bonds, options, crypto, all in one place. They'll even give you an uncapped 1% match when you transfer your investments over from another platform. If you want to build a portfolio that actually reflects your thesis, visit public.com slash sorcery. [22:41] Paid for by Public Investing. Full disclosures in the description. [22:45] Enterprise AI runs on Merge, the AI infra platform for integrations, agent tooling, and model orchestration, so your teams ship product, not plumbing. [22:54] Mistral, Dropbox, and Drada already trust merge in production. [22:58] Start building at merge.dev. Founders scale faster on Deal. Set up payroll for any country in minutes. Hire anyone anywhere. Get visas handled fast. And get back to building. Visit deal.com slash sorcery. That's D-E-E-L dot com slash sorcery. [23:17] Okay, so this is a plug for Brex, but Sorcery is sponsored by Brex. They're all about performance, spending smarter and moving faster. So with that, I like to ask, what are the metrics they use to measure the business and for success? What are the milestones for this year? Yeah, I mean, we're an at-scale business. So, you know, all the classics, top line, bottom line, gross margins. What are those numbers? Those numbers are not public. But, you know, I mean, they're healthy.

[23:47] a mature business. So we look at all those things from a traditional finance perspective. The things that are more difficult to measure, which might be even more important because those financial figures are lagging indicators of your technology and your products. We don't have good financials because we focus on the financials. We have good financials because we focus on the products, and the products are really world-class. And measuring those things are hard. There's many companies that will say, we do self-driving for X, or we do autonomy, [24:17] we do intelligence in this place and that place and how do you know objectively [24:21] either as a customer or somebody who's just visiting the office, what's actually good and what's not good. And I'll say even deeper, how do we know? [24:29] How are we not biased towards our own kind of proclivities? And those are way more nuanced. How we do benchmarking and how we think about how good the autonomous system is, and is it progressing at the right rate? And that requires a bunch of things. It requires truly subject matter experts. When we say we're a technical company, the root word of technical is technique. We have people who know these technologies extremely well. It's truly the reason the company is good, [24:59] can individually say, hey, I think this is, we're moving forward, we're not moving forward. It's not like a metric that always is up and to the right. These metrics themselves, especially when you're talking about frontier technology, can be gamed in ways that are not actually making the technology better. On the engineering side, we care a lot about productivity. And productivity is not just about intensity, although that's certainly an important factor. But there's all these technical measures that you can get very deep into that indicate, is the engineering team

[25:29] build and launch new things. And we do a ton of work in model training and AI. And there's a big metric there is like, how fast can you actually get new data and train a new model and deploy it? And you want that to be as fast as possible. [25:41] Okay. Well, earlier today you did have Marc Andreessen on the stage. You two definitely have spoken before. You're both incredibly well-read and it seems like you could talk for hours. [25:52] I've known Mark for many years. Yeah. And his retweet of your first tweet ever did go viral. Oh, yes. Yeah, yeah, yeah. So on your website, you have a couple of different sections. [26:05] You have books, you have photos and writing and more. The photos, the photography was actually really interesting. Thank you. Thank you. Not enough people give me credit for that. In some parallel universe, I would have been a photographer. [26:20] And then on the book side, I've heard some hot takes you've had on books. Yes. People that read books and don't read books. Yeah, yeah. So what's your hot take and what are your favorite books? I don't know if it's necessarily a hot take. I mean, we did kind of a Q&A like Berkshire Hathaway and the person I'm quoting is Charlie Munger. Rest in peace. But I don't think you'll meet very, very successful people that don't read a lot, that don't read a lot. [26:50] the best books to read are the old books. So over 25 years, over 50 years, because time has filtered all the kind of noise and you really get a lot of signal. And you, frankly speaking, you only get, I mean, in your life, how many books are you going to read? You're going to read maybe, most people are going to read a very small amount, but even if you're a kind of an avid reader,

[27:09] a thousand books i mean there's not many and there are lots to pick from so [27:15] just pick the really good ones because what you read does impact your view of the world. And, you know, you want to read diverse books. [27:23] that are not just kind of what, you know, your friends are telling you to read. So they really kind of challenge your existing views. And even if you come back on the other side and you still have that view, then either maybe you have a better argument against against that view. But yeah, I'm I'm a huge, huge fan of reading. I think there's also something again, a little old school, but this like long form, [27:44] deep content. I think your brain somehow processes things differently. And again, as a founder, when you read things, I also don't recommend reading tech books. I'm just one of those people as well. No airport books and things like that. And so if you read something that's really old and outside in another industry, I'll use an example, a book that's not on the book list, but is really good is My Years with General Motors by Alfred P. Sloan. The book that I have [28:14] is, which I think is even a better book, is Adventures of a White Collar Man. But my years with General Motors is this, General Motors is the company that establishes the modern corporation. [28:25] he writes a book about establishing the modern corporation. You will learn a lot about why organizations look this way from the person who really kind of, you know, made it a reality post-World War II. And that's like a very obvious one, you know, for us. But reading the, you know, autobiography of Malcolm X,

[28:45] I think that opens your mind a little bit of like, what does it mean to be American and all these things. So I highly recommend reading. You don't have to read my, you know, you don't have to read Krishna Murthy's The First and Last Freedom, but I do recommend it. And there's many others on that list that you can try out. [29:05] Peter, what are you reading? As CTO in the AI era, what are you reading? Is it just research papers? I read all kinds of things. Actually, the most recent book I read was actually Apple in China. Oh, so good. So good. Fascinating read because of the combination of geopolitics and technology and just how much investment has flowed into developing the Chinese electronics ecosystem. [29:35] We also read books as a company, as like a leadership team, and then we invite the whole company to do it. So we've read probably half a dozen books. They're work-related. They're used to reflect on our company. So we will then chunk out the book into three or four parts and then talk about those parts and then record it and distribute it to the other company. So we've done, you know, The Hard Thing About Hard Things, just plugging A16Z as much as I can. You know, we did House of Huawei, really fascinating book, super good to read. [30:05] It breaks the rule of very old, but it is definitely very, very good. We've done a bunch of this. The Netflix No Rules Rules. So I recommend, as a founder, again, I think you don't have to take all of Reed Hastings things about Netflix, but if you talk to your leadership team about they did this for compensation and they did this for recruiting, it forces you to reflect on what we're doing. And are we doing the right thing or should we now change it? Because

[30:32] Your company, it's like, think about your company. Again, I'm talking to founders here. Think about your company as like this bespoke piece of clothing that you get made. And then you grow. And then before you know it, it's kind of, and you have to constantly be retailoring your company for the scale position and type of company it is and type of products you're, and I think these books just help you reflect on like, you know, Made in America by Sam Walton, phenomenal book, excellent, excellent book. [31:02] in the late '80s. [31:03] Thank you. [31:04] Are we doing it the right way? And you will get you will get lessons there. But macro point read outside of like business and technology. You'll get even more value there. [31:13] books are good. Yes, yes. Reading good. Reading good. Yeah, yeah. All right. Thank you guys so much. Yeah, thanks for having us. I really appreciate the time and the tour and seeing everything today. Yeah, yeah. Sorry we couldn't give you any, you know, viral moments where, you know, I'd jump off. Well, next time maybe bring the mining equipment and I'll get in the really, really big truck if you can drive that around. There you go. Awesome. It'll drive itself. It'll drive itself. Yeah, exactly. [31:39] Thank you very much. Thank you. Appreciate it. Hey, it's Molly. If you enjoy our interviews, check out our newsletter, sorcery.bc, where we deliver a once a week top deals and tech headlines email and also go deeper on our podcast interviews. Subscribe to Sorcery today. And don't forget to subscribe to the podcast on YouTube, Spotify, Apple, or wherever you listen. Link in description to sign up.