Inside LA's Factories Trying To Outbuild China

Let's look at these cool machines while we talk about manufacturing for a minute. The United States still makes an incredible amount of stuff. That said, American manufacturing is in an appalling state. Well-trained craftsmen are aging out. The mom-and-pop shops that had U.S. industry humming for decades are closing. Young people don't appear to want manufacturing jobs or to take over the shops their parents ran. Meanwhile, the United States is also expensive, slow, and regulated to a point of paralysis. China can do everything we used to do, only better, cheaper, and faster. What I am describing is a crisis. Important people know this and are paying attention, but they're not doing as much as they really need to. In other words, the U.S. is quite possibly fucked. Is there any good news? Well, maybe. To the extent that the U.S. has a manufacturing revival taking place, it's happening here in Los Angeles. And to find the revival, you need to step away from Shohei and the nice restaurants. and go to the grit of places like El Segundo and Hawthorne. In these spots, you'll find loads of hardware startups trying to build new things in new ways. And the true giant of this hardware collective is Hadrian. Behold its glorious thing-making machines. Hadrian was started by Chris Power. Are you from Sydney?

Melbourne.

Melbourne, okay.

He's Australian, used to be a lawyer and an accountant, And a software salesman.

Did a bunch of e-commerce in Australia. I told the board they had 12 months to replace me. I'm either going to move to China or the US.

He came to California to repent and with the wild idea that he could be the person to bring American manufacturing back.

It was very clear that manufacturing in the US had been completely hollowed out and that a wave was going to come where it became the most important thing very quickly. Since '85 through 2020, we'd offshored 90-plus percent of all US manufacturing. It turns out that was a really bad idea. We basically gave the farm away to the CCP. Because of that, there's no one that wants those jobs. And we have this huge demographic bubble where everyone with advanced manufacturing skills is basically in their early 60s, and there's no one coming behind them to replace them. Once we got to the Ukraine-Russian war, Israel, people started to then realize like, oh, we just shipped 3 years worth of inventory to Ukraine and they blew through it in 6 weeks. And then we went back and said, hey, can we please have some more Stingers or whatever the product? And they were like, ah, maybe give us like 4 to 5 years to reramp the production. So macroeconomically, you have 2 choices. You either have to import a ton of skilled labor or you have to automate your way out of the problem. Hey everyone, this is Chris Power from Hadrian. Today, just a year after our Series C-2, I'm happy to announce our Series D. $1.37 billion in new capital at a $7.87 billion valuation to reindustrialize America.

Hadrian has raised close to $2 billion, which is incredible. Chris turned up in California in his 20s and knew no one, but people still bought into his very big idea. Good on you, Mr. Power. The point of Hadrian is to modernize and automate the work done in the past by the mom-and-pop manufacturing shops. Hadrian wants to make lots of stuff for defense, aerospace, and other industrial customers, and to do it fast, cheap, and consistently. It tries to accomplish all of this through clever software and by controlling these tools of production.

The industry operates basically on PDFs full of manufacturing hieroglyphics. Just to read one of those, you need, you know, 10 years worth of experience. So we built a lot of software to basically extract that data, convert it to a digital format. Most of the effort in advanced manufacturing is then programming the routines that the machines run.

Adrian is trying to add repeatable rigor to making complex parts. The mom-and-pop shops were very skilled, bespoke part makers. Lots of their knowledge stayed in-house inside their shops. Hadrian, however, keeps track of every part and every process that went into making it, and then refines all of this so that you can produce objects again and again with the touch of a button.

For every manufacturing domain, we hire the best experts in the country. We put them on the software engineering team, and we get that automation up to, say, 60 to 70% efficiency in each of these, you know, 100-hour tasks. Okay, we launch it and it becomes 20 to 30 hours, but then that curve quickly comes down. And through that automation, you also lower the bar of the skills you need. What we are really good at is hiring the best people and the technical ability and getting their brains into the software. Through our software platform Opus, we can train people in 30 to 40 days instead of a decade.

Are there tons of people on Hadrian's factory floor? No, there are not. If it's successful, it will not be producing jobs at the scale of an industrial giant from the 1950s. And that's obviously on purpose. Automation is king here. That said, Hadrian is pushing the idea that it can provide good-paying manufacturing jobs to people with minimal manufacturing experience.

What's up, man?

Case in point, Tony.

Where are you from?

A little bit of everywhere.

Yeah?

Yeah.

Where'd you grow up?

London for a little while. Caribbean.

Oh cool.

For most of my life.

I heard you were a diver.

Yeah, dive instructor.

How did you end up here from diving?

Oh my goodness. Got married, came to America. You know, my wife's like, try something different.

Yeah.

Let's mix it up. And I heard about aerospace. All the big names are here in California. Accidentally found this place and here I am.

Okay.

All right, so you had done like no manufacturing?

I didn't even know that machines like this existed. Before I came to H3. I came down those stairs for my interview. I'm like, what the hell am I getting myself into? But it worked out all right. I haven't broken anything yet.

That's good.

That's good.

So you want to get those gloves on, and we're going to give you a part to load. This is a drag-and-drop system, so you're going to hold down, and then you're going to drag over.

Yep.

And now the robot is going to bring it over to the setup station.

Oh, he's got eyes.

Yep, Cookie Monster. Keep Cookie Monster fed. All right, so I've unlocked the doors. So we're going to get our, our block and we're going to get it onto the vise right there. Sometimes it's a little hard to see. Don't worry, I have trouble doing it.

Is that good?

All right, that's good. All right, so we're going to get our torque. And then you're just going to tighten that until you hear a click. You're going to close your doors and then release it to the machine. And that's it.

I like the eyes.

Yeah.

Hadrian started out with a focus on making parts for things like rockets, planes, and missiles. But now it's looking to make even more of what its defense customers want, which could include building entire products from start to finish in its factories.

For our factories as a service business, if you show up with a design of a full product, we can probably do 80% of it right now. So we're at that point now, and there'll be some big announcements this year where for the last 12 months we've been launching every manufacturing category you can possibly think of, both at the component level and then the product level. So now we're Most of our contracts are like not just missile parts, but like the full missile.

Yeah. You guys have 3 factories?

We have 3 main ones.

Yeah.

Factory 2, which you see behind me, is completely max capacity. We scaled this in 18 months. Arizona, which launched December, is 5 times this size. Factory X is where we do all of our R&D because now we're building the biggest automated welding team in the country. We have an additive team. We're doing sheet metal. We're building a foundry.

This is dope, man.

Yeah, no, it's awesome.

How big is this?

This is 200,000 square feet. Yeah, so a little bit of a big playground for us to use and build out our new capabilities. A little bit bigger than the back of F2 where we were first doing it.

Zoom zoom!

Yeah.

So one of the things we're actually able to do with the automations that we build is actually capture all of the tribal knowledge of these craftsmen and artisans that have been in industry for decades and worked their way up through apprenticeships, encode that within our software and our automations, and lower the kind of barrier of entry. The operations team we have here is absolutely all-stars. They're awesome. I mean, but most of them are composed of people that have never been in the manufacturing space before. So we're able to take a brand new team member, train them up in a month, and they're cutting parts on our CNCs that you saw in the factory.

And so I think we get to like see welding?

Yeah, we do.

Hello. Hey, how's it going? What do you do at this fine company?

My name is Zach Christie. I'm a software and robotics specialist, kind of with a focus on weld automation.

Can't cross this line, right?

Yeah, correct.

What happens if we cross the line?

If you cross the line, the whole system locks down and it says you got to reset all the safety and make sure nobody's inside. So it's, it's really just there for your protection.

Okay.

So the mission today is to prove that anyone can be a machine operator. So we're going to have you actually start off this process here and make one of our I don't like how you phrase that, man.

How do you know the skills I'm bringing in?

I don't know.

No, I have no skills.

All right, well, the good news is that's as much as you need.

Okay.

Go ahead and push the green button.

Right.

It'll move around a little bit first, and then you'll probably want to mask up.

Here we go.

Get ready.

I'm welding.

You got a few more.

Some of this is like a little bit of a miracle. Like, getting it working is hard. The first 70% comes fast. The last 30% is like this enormous long tail kind of brutal problem. But that's actually fine because once you're 70% better than anybody else on time or cost or whatever, it's like super doable. What we say internally is basically like all the answers are out there. We're inventing a lot of new software and processes, but people already know how to do this and do it incredibly well. You've just got to scale it. And the way to scale it is with software and software-defined processes.

All right. Nice.

Better than I could have done myself.

I'm a natural. Do we get to keep it?

If you'd like. Well, actually, no, I have something better for you guys. Made a little gift for you guys to take home.

Let's go.

There we go.

I'm here.

We must now interrupt this episode for a lighthearted segment we're calling Liberty or Subjugation.

This is the ultimate test.

The hardware startups of El Segundo have received a ton of attention for their patriotism, faith, and energy. But we wondered, do they really have what it takes to bring America back Do they have the Chinese work ethic?

It's a Saturday and it's Valentine's Day, so we've set the bar high. So we set the bar high.

And so we went through El Segundo to see who was willing to work on a weekend and to rate the startups with bald eagles.

One to ten bald eagle scale. Empty parking lot is zero bald eagles.

I'm sad to say that most of our victims failed the test.

Quantum Space is empty. They've got zero bald eagles. Oh, this is Radiant. This is the future of our energy industry. Empty. Is that a Prius? Oh, God. There's a security guy coming.

Security guy?

Yeah, this is not a good spot.

Let's go.

Maybe one bald eagle for the security guard. Good for him.

Sift.

They make software for aerospace, autonomous vehicles. Now, no lights on, no humans, no freedom. That's zero bald eagles. Durin Mining Technology. Okay. Durin builds and operates drill rigs To accelerate mineral discovery. But as you can see, it is empty. 2 Teslas is bare minimum. This is 1 bald eagle. Right now in China, there's 100 million 25-year-olds working at the competitor to this company. And they're— these guys already have Teslas, as if they've already made it. It's pathetic.

Some had a mid, Icarus.

All right, let's see what Icarus does. Autonomous aircraft that can stay up for a long time. Rapid aerial intelligence. They've got a pretty full parking lot. Icarus, A+. Look, look at the fucking bars, got plenty of people.

We will give Valor Atomics a retroactive pat on the back here. We learned later that most of the staff spent the weekend flying nuclear reactor parts to Utah. Anything called Operation Windlord gets all the bald eagles.

Rainmaker! I knew it!

The folks at Rainmaker did well. Great American Car out front, dudes in the workshop.

Look at that.

All right, we're going in.

Do you hear it? Do you hear it, guys? But they were not happy to see us.

If you guys don't have an appointment, can I ask you to step outside?

Yeah, yeah, yeah, yeah. Well, you guys keep freedom alive.

Yes, sir.

Okay.

Gotta protect that intellectual property, boys.

What do we give them? I mean, they got people grinding. They've got a Corvette in the driveway. I'm gonna go 8 bald eagles. 8.2 for the Corvette.

And the real stars of our Saturday raid were Shinkei Systems, which makes a robot to kill fish. There they were, beavering away all Saturday morning. Although, we did make them do this so that we could give you this episode. Click, click, click. Here ends this edition of Liberty or subjugation. But beware, America, we could be visiting at any time.

Are they not here anymore? I don't know.

This is obviously not Hadrian. It is Ethereal Machines in India. They're also in the metal cutting business, except they make their own CNC machines from scratch, 5-axis machines. This is part of a wave of the rise of hard tech and manufacturing in India. We're going to have tons of stories on this coming to you on this YouTube channel. Thank you for liking, subscribing, and watching. And thank you so much to Brex for making these journeys and episodes possible. Brex fixed our finances. They'll fix your finances too.

Thank you so much for watching. And go.

Where Hadrian was all about perfecting things we already know how to do well, Machina Labs is all about teaching robots new tricks.

All right, welcome to Robot Wonderland.

Machina was founded by Ed Mayer. Like Chris, he's a foreigner.

So I am Iranian.

Ed grew up in Iran and then came to the US and ended up working for places like SpaceX and Relativity Space. In 2019, Ed started Machina with the idea that he could train robots to work with sheet metal and shape and form the metal into parts for use across American industry. The company began with what it calls RoboCraftsmen cells.

So this is a cell we call RoboCraftsman. It's 2 robotic arms working together. The core operation is sheet forming. So traditionally, if you want to form a sheet metal part, you usually need a giant 4-story stamping press, big dies depending on the size of the part— could be the size of a car— and move them with cranes, huge, huge equipment. The molds look like male and female, sheet in between them, you stamp them, you get the shape. So these robots are doing it the way a craftsman Craftsman would do it, right? A craftsman would, you know, almost hammer it into shape. The best analogy is like how a potter forms clay bowl with their fingers. They pinch it and then deform it into shape. People, when they look at something like stamping, they look at the press just did a clunk and then you have a part. Okay, but in order to get that clunk, like there's so much investment in. space and also, like, you know, equipment ahead of it. If your factory wants to go from Model 1 to Model 2, it's a matter of days just reprogramming it. It's fully software-defined to go from one design to the other.

So this is Factory 2, or—

This is Factory 2. We call it Mach-E, and this is where we're gonna ramp up the production of the cells. Also, most of the systems you saw in the other building were our first version.

Yeah.

Here we actually build a more robust, ruggedized version of the cell. We call it V2. So, and the goal is in this facility we can have production scale up, up to 25 to 30 a year of these platforms that we can make. And the new versions are very different than the original, right? They are portable. Here, as you can see, the frame of a robot that's just getting assembled. Yeah, these 2 are actually 2 containers, right? So the back of the container kind of folds up.

Oh, it all folds It folds up like that?

Yes, and then you get 2 20-foot containers.

It's basically a shipping container?

Basically a shipping container.

Wow, okay.

Yeah, so we have shipped this to a war department.

Okay.

But also it's like for our own usage. So we're setting up our 3rd facility now. So the idea is that you can set up a factory within a few weeks, and that same factory can start manufacturing any kind of part really fast.

Okay.

This cell turns into a container, gets shipped, it lands on any warehouse floor. As long as there's power, it self- calibrate, and then within the same day can start making parts.

So you've got manufacturing in the field.

Right, immediately set up. Like if tomorrow we have, let's say, conflict in Pacific, we can set up a facility in Philippines, in South Korea, in Japan in a very short amount of time. I actually deployed to Arkansas a few months ago, and we were in this area where they didn't have industrial power. So one of our partners is Toyota.

Yeah.

And we actually got a truck from them that run a compact, you know, hydrogen generator on its back.

Yeah.

And that could run it for days, for a month. The funny thing was the system was making parts for that Toyota car. But tool change is gonna happen here, so this is where the tools are.

Right here? Yeah.

Our quest today was to build a drone.

Slightly sexual, Ed.

So it dropped its It's forming hands. Now it's picking up its eyes. And now it's gonna go and scan the formed drone. It's basically a laser profiler. And then we combine that data with where the joints are and the kinematic information about the robot to create a 3D map.

Okay.

If you come here, you will see that that 3D data is being captured.

Oh yeah, okay.

So now it's actually going over the part.

Oh yeah, yeah, yeah.

The same robotic cell that formed them can scan it, map the cutting path.

Okay.

Then do the cutting. So you get the part coming out of the sheet, and then we'll take it for welding.

These machines are trying to replicate skills that tradespeople learn over decades. To do this, the robots make a part again and again and have AI software and humans watch over them. and hone the techniques.

We're gonna bring the human operator in the loop so the human operator can give it a set of instructions, then look at the results, make modifications to improve. Oh, it tore. Okay, these are the changes you wanna make. But then over time, as we are gathering more and more data, we can kind of like reduce the number of trials that human operator can do it and eventually, hopefully, like reduce any involvement that the human operator has.

Makano is competing against massive industrial machines that can stamp out tons of parts fast and cheap. Ed's argument, though, is that the robots make it easier to prototype and to make design changes on the fly. And he believes that building custom objects on demand will become a big thing.

In early days, we received the part from SpaceX related to Starship, and then we had to do 25 trials to get the recipe right. Now we are averaging 4 or 5 times.

This is reduced.

The goal is like zero-shot it, right? Right off the bat, you get it right.

I was even allowed to participate in the drone-making process. Sorry.

There you go.

There you go.

Yeah, no, you got it.

Yeah, that looks good.

Like Hadrian, Machina is trying to teach its robots new skills to make more and more of what American industry needs.

It's gonna be a factory that looks like a data center. Our 3rd facility, we're planning to deploy 50 of these Robocraftsmen in it. You can program each of them to do different operations, different types of parts, and then become a fully software-defined factory. So right now we have productionized forming, trimming, and scanning, which is basically QC. Our next operation we're working on right now is heat treatment. We are looking at bending, hemming, forging, welding, and assembly. Then you pretty much can do any sheet metal basis metal structure.

Okay, some final thoughts while Ed and I weld together in his garage. Like men.

Jesus. Yeah, there you go.

You should know that Hadrian is quite defense industry dependent at the moment.

Then you can start bending it. Yeah, bend it.

And you should know that defense spending whims change with the political winds.

Oh, shit.

Yeah.

There are also people who think Hadrian will struggle greatly to automate these manufacturing processes and that it's been overhyped. As for Machina, well, it has much to prove as well. Lots of people aren't convinced that the robots can outflank the mass production machines of today, and they argue that the bots will forever be too slow and too expensive. And so Machina still has to find its place.

We've made a Tesseract.

Core memory.

Yes, yes.

Ultimately, the The stakes are incredibly high, and these are people who are actually trying to fix the problems.

The thing that's clear to me is everything's gotten bigger. There's more robots. You guys have made all this progress. Then when I play in my head of like trying to compete against China, man, it feels like we need tens of thousands of all these machines cranking away. Why do we not have 2,000 of yourselves in a giant building today. Yeah.

Well, one difference we have from China is that the government cannot as uniformly make an action. But, you know, I guess one beauty of the United States is you need to find your way in the market, right? So I think a lot of the work that we've been doing in the past few years was finding the right application that can make us money right off the bat so we don't have to rely on giant government subsidies to scale this, which is, I think, long-term more resilient. Well, yeah, the downside is the speed is slightly slower.

Arizona's up and running. It feels big right now. The other one we'll announce later this year is going to be 2 million square foot. It's sort of like China scale. It's like, it's a drop in the ocean. It's tough and we're so far behind. But the choice is like you either give up or you do it. And then, yeah, the other thing is it's going to take 10 to 20 years. But like, you know.

So you think you just, you have to get started somewhere and you push it as far as you can?

Yeah, yeah, yeah.

Don't think too hard.

Yeah.