Defence startup: AI shrinks design time to 20 minutes (!!)

Christian Mochen: I mean, when we’re on calls with customers, we’re able to generate new motor designs for them while we’re on our first call with them, which is unheard of in the space.

John Koetsier: Can AI take two months of engineering a physical product down to 20 minutes? According to the CEO of a company just emerging now from stealth, absolutely. Today, we’re chatting with Christian Mochen, co-founder and CEO of Atlas Motion. Welcome, Christian. How are you doing?

Christian Mochen: I’m doing fantastic. It’s a great morning over here, John. I appreciate you having me on.

John Koetsier: Awesome. It is my Friday afternoon, and I’m spending it with you, so hopefully it’s going to be good. Let’s start at the top. What does your company build? What do you do? What do you make?

Christian Mochen: Yeah. From a macro level, we make motors and actuators for robots, drones, autonomous platforms, and things like that. But we leverage a pretty sophisticated manufacturing process to do that, which I think sets us apart pretty well.

John Koetsier: We’re going to get into that, and we’re going to get into how you use AI in the process and take that time down. I know a little bit about engineering. I know some engineers. My son’s an engineer, and they spend months building new products, prototyping them, and then releasing them. You’ve managed to find a way to really cut down on that.

Before we jump into that conversation, talk about why you’re building motors and actuators and why that’s critically important today, right now.

Christian Mochen: Yeah. When I get into that question, I like to frame the company from a historical perspective. In the 1940s, the U.S. and its allies manufactured their way to a win in World War II and essentially became a global hegemon.

In the late ’80s and early ’90s, China very cleverly started subsidizing critical industries like rare-earth mineral refinement, mining, and manufacturing, from your bare-state, lowest-level minerals all the way through to the end-state manufactured product, like your DJI drones. It essentially decapitated our ability to create those critical components for ourselves through heavy subsidization.

Now, in the age of physical AI, we believe that the next critical problem for our industry will be bringing that capability back to us and maintaining that sovereignty as we move into the next phase of these future technologies. I think it’s critically important.

Our team has deep experience in the defense-tech industry. I started at Tesla, went to Shield AI, and then eventually Mock. My co-founder, Carlo, whom I met at Shield AI, traveled with me to Mock. Then there’s Tom Barron, who was VP of strategic initiatives at Mock. We all saw these problems happening in real time, and we all saw how difficult it was to source these components in the supply chain. It was sort of a moral obligation for us.

John Koetsier: Everybody’s seen what’s going on in Ukraine. Everybody’s seeing what’s going on in Iran and the Strait of Hormuz, where you’ve got asymmetric warfare and $300 drones taking things out—or requiring multimillion-dollar missiles to take them out. That’s unsustainable for any country. I don’t care how rich you are.

That’s super interesting. That’s what you’re building. You’re going to supply industry in the United States with that.

One of the most interesting things about what you’re doing is how you’re doing it. You’re talking about taking a multiweek, multimonth engineering process of designing a new motor or a new actuator and reducing that drastically. Talk about what that looks like, how that works, and how you maintain quality through that.

Christian Mochen: The first thing I like to highlight about our platform, which is called Vector, is that AI isn’t replacing engineering. It’s simply orchestrating a lot of critical decisions across the engineering toolpath very, very fast.

Typically, what you’d see in this sort of process is a bunch of cross-functional engineering teams working in their own lanes and avenues. For a motor, you might have an electromagnetism person and a mechanical engineer. You have your supply-chain, manufacturing, and logistics teams working around the clock to make sure you can actually take this design, move it into production, manufacture it, and deliver it at scale.

Our platform essentially understands the context and input of the design that you’re trying to build. It brings in a mesh environment, simulates all the platforms—or all the tools—that you need to test that platform theoretically, and then builds out all of the manufacturing packages that you need to execute the build process.

It does that all in five to 10 minutes. There are some very small tweaks you need to make on the back end, but right now we’re seeing that we’re within one percent—or, I’m sorry, at 99 percent confidence—of what we’re actually seeing when we output hardware on the floor.

It’s a pretty robust platform. We plan on going much further with it. The inspiration for it was actually from Tesla. When I was there, we had a platform called Odin that orchestrated the self-testing procedures for the cars throughout the production line. We wanted to go a step further with that.

Typically, the biggest primitives you have in manufacturing are your equipment and your tooling. They’re like black boxes of capability that you spread throughout your manufacturing floor. We wanted to go a step further than that. We wanted to expose physical state as the primitive.

What that meant was that we swapped compute throughout our tooling chain in our manufacturing process and then embedded this orchestrating system, Vector, which understands the context of the design. It drives that information through our production process and actually orchestrates what we’re doing on the manufacturing floor. Then that data feeds back into the system and makes our model smarter over time.

John Koetsier: Interesting. What does an engineer’s job look like here? You’re still designing a product. It still needs to do this and have this capability. What does that look like for an engineer working with a system?

Christian Mochen: Internally, we don’t really call ourselves engineers. I mean, we’re all engineers by trade, but what we really call ourselves is architects.

Our engineers are very systems-level thinkers. They design from first principles, but the hard part of the design is covered by our software. What they’re really doing is architecting the inputs to create a really robust product on the back end.

They’re more like product architects, in a sense—full-stack product architects. They know a little bit about everything: electromechanical engineering, mechanical engineering, software, the whole shebang. They package that all into a delivery for our product.

John Koetsier: This is crazy because this mirrors what I’m hearing elsewhere in other industries. Typically, these industries are not producing a physical output. People are renaming positions as builders, creators, or whatever, and people are orchestrating.

You are the CEO of your own mini-company, and you’re directing agents and AIs to do different things. I didn’t expect to see it today, at least in terms of physical products, and especially high-performance, mission-critical physical products. That’s another step.

Christian Mochen: Yeah. It’s pretty remarkable what we’re able to do with this platform right now. From a bare-bones perspective, we have this saying internally: Building a motor is like making sushi.

Typically, making sushi is not the most complex thing in the world, and it doesn’t have that many ingredients. But when you have a bad piece of sushi, it’s very, very obvious. What our software does is allow us to make really good sushi really fast, as we like to say.

John Koetsier: I’m going to read something you sent me. We talked already about converting a new spec into a finished design in 20 minutes, compared to two months. Then you take that design and scale it in production in six weeks. The industry standard here is more like six months or something like that, right?

That is a game changer because you can have more product turns, more innovation, and more product iterations. You can learn more quickly and get better faster, correct?

Christian Mochen: Yeah, exactly.

John Koetsier: Interesting. You’re making 10,000 motors a month right now. Can you talk about your customers?

Christian Mochen: Right now, I’m not in a position where I can discuss many of our customers, so I’ll probably keep them under wraps.

John Koetsier: But you’re growing to 40,000 a month in December. Is that correct?

Christian Mochen: That’s correct, yes.

John Koetsier: Interesting. You still can’t talk about those customers either?

Christian Mochen: No, I can’t.

John Koetsier: Okay. Clearly, top-secret customers, probably in defense and other critical areas, and that’s super interesting.

Now, you are coming out of stealth, and you’re announcing a raise as well. Talk about that.

Christian Mochen: Yes, we are coming out of stealth. We’ve raised $11.5 million, led by Greycroft, with participation from Also Capital, Sunflower, Banner VC, ENEA, and a couple of others, along with some pretty great angels: Jay Malik from Amca and Scott Sanders, CGO for Terra. I’m very, very proud of the crew we’ve put together here.

John Koetsier: It’s not too many times that a company comes out of stealth and is already producing 10,000 motors per month. You’ve been going for a while. You’ve obviously had a reasonable seed round, and you’ve been growing significantly already. That’s impressive. You’re scaling to 40,000.

Christian Mochen: I would say “a while” is relative. We incorporated in February, so it’s been a little less than six months. We delivered our first motor in fewer than 60 days and have just been scaling super aggressively.

John Koetsier: Hardware is the new software. You never heard about this from hardware companies before. Hardware companies would get a little seed round, or maybe even a Series A, and disappear for three years. Then they’d pop their heads up and say, “Hey, now we have a product.”

Software companies were the ones quickly iterating, quickly building things out, building in public, and doing all those other things. Hardware seems to be the new software. I guess AI is the difference.

Christian Mochen: I love it. I love to see this shift. I think the biggest bottleneck is going to be physical systems and the components that make that happen. I love to see hardware moving at the speed of software.

I agree. I think it’s going to be the most critical part moving forward over the next five to 10 years.

John Koetsier: Super interesting. Will your actuators go into other types of robots? You talk about drones, but what about humanoid robots, quadrupeds, or anything like that?

Christian Mochen: Absolutely. We are making instruments of motion, and we want that motion to unlock everything under the sun. We’re going to be supplying every industry we can think of: industrial robotics, humanoid robotics, general-purpose robotics, specialized robotics—the whole shebang.

We want to be a globally competitive business. It’s a big reason why we set up our company the way we did. We want it to move extremely fast and extremely aggressively. We’re ruthlessly pragmatic. We want to capture the market as fast as possible, and we want to be globally competitive.

We don’t want to rely on government subsidies to create an effective business model. We want to sell into India, Poland, and Africa. We do have customers there, where China is still a factor in the market. That’s a big reason why we set up our company the way we did.

John Koetsier: If you’re growing the way you’re talking about here, and you have the iteration capability you’ve mentioned, you’re probably looking at some significant major customers. You’re probably looking at their exact specs and saying, under certain circumstances, “We’ll design to your exact specs.” Is that correct?

Christian Mochen: Yeah, exactly. The speed of delivery and scale is a big differentiator for us in the space. But really, we like to think of the business in terms of LTV to CAC.

Typically, in the supplier space, LTV to CAC is absolutely horrendous. It forces you to operate off a catalog motor SKU model, specifically in our industry. A lot of hardware is like this as well. The effective cost of changeover and downstream operations is just not worth opening up a wide range of highly variable products entering your production process.

For us, with the speed of our software, the standardization of all our raw-material inputs, and how we’ve set up our manufacturing process, changeover when it comes to new designs flowing into our manufacturing operations is relatively small.

We’re driving the effective cost of iteration as close to zero as possible, which allows us to operate in ways that typical component suppliers can’t. That’s why you’re seeing companies in Africa, India, and Poland actually procure from us when they could procure from China. We’re able to move faster than them, and we’re able to give them more flexibility.

When we’re on calls with customers, we’re able to generate new motor designs for them while we’re on our first call with them, which is unheard of in the space.

John Koetsier: Yes, it is unheard of in the space. Wow. You’re doing manufacturing in the Philippines, correct?

Christian Mochen: That’s correct—large-scale manufacturing. We have two arms. We have our engineering and prototyping arm in Long Beach, California, and then we have our mass-scale production facility in the Philippines.

John Koetsier: Okay, interesting. Very, very cool. You could still be hit by tariffs or something like that when bringing products domestically into the United States.

Christian Mochen: Potentially. We’re putting a good team together to account for that. Luckily, the Philippines is signing the Pact Silica. They’re the longest treaty ally of the United States, and we have a very good relationship.

We’re a PESA-accredited business, so we’re registered for import and export. We have very good benefits for operating in the area, so things for us aren’t too bad out there.

John Koetsier: Very, very cool. Thanks so much for your time, and congratulations on the raise. We’ll see what you do in the future. It’s a fascinating way to build a business, especially a hardware business.

Custom manufacturing for different people and driving the cost of iteration as close as possible to zero—hardware seems like the new software.

Christian Mochen: Yeah, I agree. Thanks, John. It was a pleasure. I appreciate your time.