Atomic Machines emerges from 6 years of stealth with $250M raised to build matter compilers for micro-machines

Speaker 1: Black Screen. We are working on Jeff. Here he is.

Speaker 2: Welcome to the show. How are you doing, Jeff?

Speaker 14: Good. How are doing?

Speaker 2: We're good. Thank you so much for coming on the show. Since this your first time, I would love for you to kick it off with an introduction of yourself and the company.

Speaker 14: Oh, great. Yeah. So so let's see. The the nutshell version of me is, you know, I'm a computer scientist who went very multi... Multidisciplinary over my career. Mhmm. My my first manager out of college was Jeff Bezos. He was my manager at D. E. Shaw and Company in New York and I followed him to Amazon and so I was engineer number 10 at Amazon and it really built the, you know, sort of led the charge to build the back end of Amazon, kind all the fulfillment aspects of it. And then went over and I joined us team and ran the consumer side for about four years and built Amazon Prime amongst other things. So I'm I'm the one to blame. So if you're building find yourself spending too much.

Speaker 2: The matter compiler from the diamond age. Explain for everyone Yes. What that is and how you're actually going about building it.

Speaker 14: Yes. The atomic... Atomic machines is... It just come out of stealth after six years. We're like turning on mushrooms.

Speaker 9: Overlooking... How much did

Speaker 2: you raise in the process of coming out of stealth?

Speaker 14: 250,000,000. A

Speaker 1: very a very humble amount for your pedigree, I will say. But

Speaker 14: Thank you.

Speaker 4: It's a

Speaker 14: good start. Yes. It does it does seem small these days. But we've made it work well, so yeah. What the Matter Compiler does, it's new kind of manufacturing technology, a new manufacturing system. It's new because what it does is it takes, digital code and produces a complete working micro machine out the other side. And that's they're just this is the technical name for this is a combinatorial constructor, so David Deutsch reference. But, you know, this type of technology doesn't exist today. You know, the closest thing would be something like a three d printer, which can make different shapes in material. But this is, you know, complete working micro machines with moving parts, you know, that that, you know, to serve a purpose. You you you basically describe to an AI.

Speaker 2: Yeah. So this is building the gray goo machine from the book, don't build the gray goo machine?

Speaker 14: We're not quite at gray goo yet. Okay. You know? So, yeah, so this is, you know, it's actually, there's Neill Stephenson who was obviously just on a little while ago is, you know, is one of... Is like my favorite author. And we... The matter compiler is actually attributed to, you know, the matter compiler in the diamond age. But I would say we're closer to the manufacturing technology in Snow Crash. Okay. Yeah. So this is like the auto factories of Snow Crash, know, we're we're still working with material, you know, kind of like, you know, macro materials. We're building micro machines, so features in the microns. But we're not working yet with atomic scale. That's Sure. That's part of our mission.

Speaker 2: But the actual machine might be a centimeter?

Speaker 14: Yeah. Millimeters in millimeters in scale.

Speaker 2: Yeah. That's small.

Speaker 14: Our first device is is called PrimeSwitch. It's a it's a new class of power relay, and that's that's a 9.5 millimeter diameter by three millimeter thick device, and it can carry a 150 amps Yeah. And open in fifty microseconds. So, yeah, new new thing for the world.

Speaker 2: What is the TAM on these types of machines? Like, seems like if they're used in data centers, they're used probably all over the place. But what what, like, what is your go to market? Are you going to conferences for people that need these particular machines at the scale? It feels like it's very diffuse. Diffuse is probably like a piece in the supply chain of, you know, so many everything from consumer electronics to medical equipment to data center equipment, etcetera.

Speaker 14: Yeah. Yeah. It's true. Like, what we did is we essentially planted a flag in a relatively unserved space, right? Sure. Micro machines, you know, it's a and this is the reason we started here is because it's out of reach of conventional manufacturing. You just can't use traditional kind of macro scale manufacturing technologies because the physics kind of fails you as you go down into the low digit microns. And so what you have left in to make something, anything at any kind of volume at that scale is semiconductor manufacturing. And semiconductor manufacturing was invented to make one kind of machine, right? It's the semiconductor transistor and circuits built around that. So that leaves the whole world of micro machines, which is you can think of as kind of several different domains of physics crossed with kind of a couple of different purposes. It leaves that whole space unserved. And so we've built the first manufacturing system that can actually really fully address that space. But we also went beyond that and we built it in a way that is a combinatorial constructor so that what we're hoping to trigger is a massive kind of innovation explosion at that scale. And you mentioned some of the kind of domains applications. I this is a negative space, so it's hard to put a TAM on it. I think it's extremely large, of course. But things like robotics, micro robotics that can go into difficult environments, including the human body. Medical micro robotics is something people have talked forever, right? It's like, in the early 2000 people were like, it's here, but it's not really here. Turns out you just can't build those things. And like one example of that is I was actually talking to Eddie Chang, who's the head of neurosurgery at UCSF. He said, what we need for a watershed in neurosurgery is like a millimeter scale micro robot that we can insert into the through the cranium into the brain and then steer it to its destination. Sounds It kind of creepy, but like the imagine if you could just if you could carry a little drug payload to any place in the brain. Today, you're limited to you make a hole in the cranium and then whatever you can reach with a straight line tool is what you can get to. And that limits the areas you can operate on. And that's a really simple idea, you know, and it's but like there's just never been a manufacturing technology to make that device. And so, you know, some medical applications, know, little tiny, you know, grain sort of size of a piece of rice kind of dosing pumps, these kinds of things. That's the stuff we'll be able to build. And I'm super excited about medical applications.

Speaker 2: Yeah. Yeah. Very exciting.

Speaker 1: One of the... Yeah. A very fitting day to have you on the same day as Neil. When you were... Prior to six years ago, post Amazon, did you just decide I'm gonna read a lot of science fiction and figure out what to build? Like, where where was the... I was asking Neil about how he feels, like, what kind of weight he feels in terms of writing science fiction because Yeah. For better or worse, you sort of come up with ideas that are plausible, believable, but but far out and and entertaining, and and yet then generations of of humans read that and imagine what's possible, and then Yeah. And then it becomes possible because somebody imagined it in a in a sort of fictional context.

Speaker 14: Yeah. Right on. Yeah. No. I mean, I was definitely influenced by by science fiction. I mean, it started for me at five years old with Star Trek. In fact, there's an Enterprise right behind me. That's the original the original series, which you guys probably don't remember. But, you know, Captain Kirk and Mr. Spock and those guys. Yeah. Basically, you know, this idea that you you could like, you know, the the... A purpose for existing exploration and learning and curiosity was like, it blew my mind. And the idea that you could, you know, you have the ability, our species has the ability to like edit our future, you know, to build for problems we can't yet anticipate, you know, to create new knowledge and to build new types of technologies, new physical things. And I just wanted to be part of that. Like, from from when I was a kid, I wanted to I wanted to get a push on the frontier of what humanity can do. And so, you know, I've done that in different ways. Like, you know, I was, know, at Uber and another thing there is like, by Uber, you know, that's the elevator aircraft. Yeah. So, you know, the the the EV tall world. I mean, so, you know, I've... That's a... That's been a passion of mine for a long time. And I developed a thesis around this that I call the human command of matter. And it's... And it's... You know, it led to the creation of atomic machines and it led to our very ambitious mission, which is on demand universal command of matter. We... To get to get there, you're really at the point that Neil envisioned in the diamond age where it might not be done. If there's the feed and the seed, those are two different versions of that world. I hope it's the seed, not the feed. But where you can literally create any sort of physically stable configuration of atoms that quantum mechanics allows. Right? That's what we want. That's when we're with that, I'll probably take a vacation.

Speaker 2: I love it.

Speaker 1: What's a more... Leave us with a sort of maybe a more cute or trivial use case besides you know, medical applications? Like, I'm imagining, could we get mosquito sized robots that defend your home from mosquitoes? Like, how how far can can you Yeah. Take

Speaker 14: Sure. I mean, you know, there's kinds of things like that, you know. And I and our our approach is going to be to open up the matter compiler as a foundry. And so we'll let the world invent what those devices should be. If it's mosquito killing machines or it's, you know, graphene on copper super capacitors or it's, you know, you know, a a cooling... Two phase cooling loop for a GPU. Right? Something that you just can't, know, you can like bond to the chip and put inside the package. I mean, these are all these are all kinds of applications. But, you know, our goal is to get to the point where, you know, you you interact with an AI just the way you would with, like, Cloud Code or, you know, Codex. Right? You you interact with an AI, and the AI is generating output tokens, some of which are physical, you know. So you... It's... Know, and and the reason, like, you know, Yeah. If you think about it, the the reason you can do cloud code is because beneath beneath this... You have this new layer of intelligence which lets you vibe code. Right? You can you can like talk to the intelligence. The intelligence will map your messy natural language into a formal grammar of a compiler. And you have a compiler with a formal grammar. There isn't something like that for Adams. Right? You can't do the same thing. You can talk... You might even... Even if you... You know, kind of, presume that there's a a clog code two point o that that, you know, now for atoms, and it says, know, hey, you know, go ahead and design anything you want, any machine. There's no formal language to express the machine in to then compile it into a physical thing. Mhmm. So you then have to go figure out how you're gonna manufacture that thing. And that's that's the kind of the the world we live in today. Right? It's it's like the manufacturing line is a kind of per device thing or per kind of device thing. You have to build kind of from scratch and figure it out and tweak it and tune it over years. You know, that's the game we're trying to change. So that way, there is a formal grammar that you can express, you know, the the thing you're trying to build. In our world, it's called a process graph. This is real. And if you can express the correct sentence in that formal grammar, you can get the machine out the other side. So that's a very hard thing to build. That's why we went into a hole for six years.

Speaker 1: Yeah. Well... Yeah. It's fascinating. I I I feel like we're going to look back in fifty years and look and and and look at, you know, pre 2030 and be like, back then, we could only build big machines. The only machines you could build, you could Car. Were massive. Yeah. The car, the plane, etcetera. Right. Very, very fascinating. I'm sure you'll be back on a lot, and we'd love to have you.

Speaker 2: Congrats on the fundraising coming out of Stealth.

Speaker 1: Yeah. Thanks.

Speaker 14: To meet you guys so much.

Speaker 2: We'll talk

Speaker 14: to Yeah. You

Speaker 1: Let's do it again soon.

Speaker 4: Have a

Speaker 14: good one.

Speaker 1: Take care. Cheers.

Speaker 2: Up next, we have Eli Manning and Michael Waldron, the executive producers of Chad Powers and we have some special I know