Saronic的4位联合创始人谈如何打造一家92.5亿美元造船公司｜第三部分

Being successful in real space, doing hard tech, building real things, and manufacturing with real, tangible items is going to be the moat of the future.

I think a lot of commercial boats target human beings to take maybe 0.4 to 0.6 Gs. Our boats have seen accelerations north of 20 Gs before. A human would not take that. It would be very dangerous for a human to be on that if it were going that fast.

The technology that we're creating is having an impact on real people and real operations. I'm looking forward to shaking the hands of those 2 Apache pilots.

What we're building at Saronic will not only provide more capabilities faster, but it will save the taxpayer hundreds of billions of dollars by driving down costs as well.

To set the table, we talked a bit about your stats when we started the last walking tour. If you haven't watched it, you should watch it. You're now at over a $9 billion valuation. You raised $2.6 billion. You've just announced—and this was a little debate about whether this would still be fresh in a month; I think it will be—but Port Alpha.

That's right.

The $3 billion Port Alpha.

Yeah.

So, to bring all that into focus against the backdrop of what's going on in America, I'd love for you to lay out the stats that we're working with. Palantir CTO Shyam just talked about this, but we're ratioed by China in shipbuilding.

1. The China Shipbuilding Gap

It's actually pretty crazy. China outbuilds the United States 230 to 1. The naval fleet that we have today is actually shrinking. There are right around 290 ships in our naval fleet. In 2018, Congress actually set a statutory minimum of 355 ships. That was 8 years ago, and we're going in the wrong direction.

In 2024, the Navy actually looked at it and said, “We don't actually need 355. We need 381.” And these are just manned ships, let alone all the unmanned ships that we're going to need in our fleet.

You see the president's budget request for $1.5 trillion. They put out a maritime executive order and the Maritime Action Plan specifically around shipbuilding. You have the SHIPS Act, which is in Congress and focused on the shipbuilding industry.

There's a lot of focus on defense technology, the industrial base, and shipbuilding, and things are changing. But when you look at the budget today, let's look at the Department of War's budget. We talked about the request for $1.5 trillion, and the budget is today right around $1 trillion for 2026. One percent—1%—went toward autonomy. That just gives you a sense of how far we actually have to go.

In terms of the bureaucracy and what needs to change, you're getting a lot of push from the top down, from leadership and from the administration, which I just highlighted, and from Capitol Hill to invest in things that are not only much more economical, but also scalable.

When you talk about ships, one of the biggest examples is aircraft carriers. They take over 10 years to make 1, and that number increases by the day because the program is just later and later and later, and the cost keeps going up and up and up.

I was actually quoting $13 billion for an aircraft carrier, and I was at an event where the president spoke just last week, and he quoted $19 billion. I was like, “What is happening to the budget here?”

So you have a lot of focus from Congress because there's a 30-year shipbuilding plan in the Navy right now. We talked about going from 290 to 381 over 30 years. That'll cost $1.2 trillion if we can even execute that plan as a country in the first place.

What we're building at Saronic will not only provide more capabilities faster, but it'll save the taxpayer hundreds of billions of dollars by driving down costs as well.

I'll give you a fun stat. I didn't actually know this, and I'll laugh at myself a little bit: the Army actually has more boats than the Navy does.

We're focused on all the services: Navy, Army, Marine Corps, and Coast Guard. How can we support the Air Force in combat search and rescue and other things? We're focused on the military at large.

The Navy was our initial—and still is our largest—customer. As we're building autonomous surface vessels, that makes a ton of sense. How do we redefine what that naval fleet of the future can look like and should look like? We're figuring that out together in partnership with the Navy.

We're very focused on all the services, very focused on commercial, and also internationally. Our allies are just as important in this effort. How do we make sure that our allies have the best technology out there?

It's not just the United States that's struggling with shipbuilding. China has 57% of total global shipbuilding capacity. A lot of our allies are struggling with the same things that the United States is: the lack of ability to build ships.

If you go back in history, even just to World War II, what has typically determined naval power is who can build the most ships. It's not who can build the most exquisite or the most expensive ships, or whose ships have the most on them in terms of payloads, weapons, and all the other things. It's who can just build the most, who has the industrial base to build the most. That kind of determines naval power, and that's the whole genesis behind Saronic.

Isn't there a certain requirement in China that you have to build every boat to military grade?

I don't know. It's a very opaque market, by the way.

I would imagine.

It's very hard to get information out of there, but they're building all types of ships. The biggest discrepancy really is in the commercial market. What that commercial market lets them do, again, is take the industrial capacity from the world. They're doing that by undercutting prices.

Their entire shipbuilding industry is completely subsidized—not only through direct construction subsidies, but they subsidize the raw materials, they subsidize the labor, and they basically give these shipyards free financing.

I'll give you a commercial stat. We're talking about 230 to 1, but it's actually not as egregious when you look at just military ships. The United States launches about 5 to 10 military ships—destroyers, cruisers, aircraft carriers, straight naval vessels, warships. China is probably doing about 30, so 6 to 1 or 7 to 1, somewhere in there.

They built over 1,000 commercial ships last year. The United States built 5. I think this was in 2024: 5 large, ocean-going commercial vessels.

What China is doing is taking the capacity from the rest of the world. Our allies, South Korea and Japan—the next 2 largest shipbuilders in the world—are actually getting a lot of their modules from China on the commercial side. They're getting a lot of their labor from China.

As that capacity diminishes, if we ever actually do get into, God forbid, a conflict in the Taiwan Strait that leads to World War III or anything dramatic like that, China can outproduce the rest of the world in terms of ships. They just keep taking that capacity by undercutting the world on price in the commercial market.

Doug, welcome to Sourcery.

Thank you. I'm excited to be here.

We're on this big Saronic tour today. You're the COO. How did you get here? How did you get involved?

2. Building Saronic From Scratch

I connected with Dino in 2022. At the time, it was actually just the 2 of us. Dino was here in the Founder Residence program at 8VC.

I bounced around maritime tech for the better part of 2 decades at this point. I'm getting old. I had just come off an opportunity working on submarines and happened to connect with Dino. It turns out it really only takes a good breakfast to get me to throw in my lot with Saronic.

From there, that was history. We started the company, started hiring, started building, and started developing. Early Saronic was super fun. I don't think I appreciated how fun it was until I looked back at it.

Mm-hmm.

It was scrappy. It was very stereotypical. We were quite literally in a garage here in South Austin. Everything was urgent, everything was new, and everything was impactful. It was a special time.

Now you've scaled up to a $9.25 billion valuation. You've raised $2.6 billion in total, and you have some pretty amazing partnerships. Walk me through some of your recent partnerships. You have one with Palantir, and you have one with Path Robotics. How did those come together?

I guess some broader context: out of the working population here at Saronic, roughly 1,600 people roll up through my organization. A lot of the executional arm of the business tends to be in operations here at Saronic, and we think about that very critically.

Palantir is pretty straightforward. We're on their Foundry program. We use them for a lot of our built-in tooling, our internal manufacturing execution systems, and some of our MRP, material resource planning tools. Think of it as the backbone software that plans out the logistics of the business. A lot of that runs on Foundry.

It's actually been a surprisingly easy adoption curve. Everyone from the supply-chain analyst to somebody in government relations and public relations is in those tool sets. They're all developing, making their own dashboards, and pushing that needle forward.

Path Robotics is interesting.

So you’re probably aware that we both have this facility here in Austin. We think about the local Texas footprint as building boats, whereas ships carry boats. Boats are significantly smaller. Franklin is obviously our first foray into ships. That’s where we’ve been exploring with Path Robotics.

The real hard problem to solve there, and it’s something that I think we’ve been collaborating on with them aggressively, is how you do aluminum welding. It’s making progress, and it’s not just, can you tack a piece of aluminum together? It’s, can you actually do it in such a way that it goes and passes all of the ABS, IMO, and Coast Guard certifications associated with that work?

What has that process been like? That one is particularly interesting because it’s very tangible, too, right?

Yeah.

What have the processes of scaling this out so massively in such a short period of time been like?

Little things that seem trivial matter a lot. I have the hardware engineering organization, as well as the supply chain organization. Silly things like part numbers, right? How you think about part numbers, how they flow through the system, how they result in CAS accounting and being able to invoice—all those things are actually really important to get right.

In the context of how we think about robotics, that’s an interesting question. Here in the boats world, we’re still what I would consider high-mix, low-volume in the context of the broader world. We’re not building 100,000 or 1 million like you might see in consumer electronics. So while it’s present, it’s a little bit of a lighter touch.

The shipyard is interesting because a lot of our focus—and we can talk about this in the context of Port Alpha later—is, how do you move a lot of the final outfitting and the on-unit, so on the ship, traditional work upfront to being done in subassemblies? That’s where we lean into solutions like Path, bringing those piece parts out, doing them somewhere else, and then assembling modules into larger modules into supermodules.

If you think about that period of time when you built Corsair, built Marauder, and now you’re in production working on Port Alpha, what do the next 10 years look like?

What’s probably not widely known is that there were actually 2 products prior to Corsair, and the reason why I bring that up is because it’s important in the context of Marauder. We built a product called Spyglass. We had a really lesser-known product called Cutlass, and then obviously we landed on Corsair.

This captures how we think about problems. We start small, we build up from there, we build momentum, we build understanding, and then we go and apply it to larger, more complicated systems. Just like Spyglass was to Corsair, you can think of Marauder as the analogous opportunity to maybe a 400-foot ship or an 800-foot ship, or maybe even something larger in the Panamax size of 1,200 feet.

How do all the operations work together? How are you managing that?

My beard wasn’t always this gray, Molly. It’s a good question. The way Saronic does business, I don’t think, is typically— it’s not different from other defense contractors. We have a platform staff that does a lot of the R&D and a lot of the development work.

In parallel with that, we have what we call Mission Operations and Services. That’s the team that deals with the test cadence, runs the ranges, and deploys vehicles all across the world. They’re also our first-line customer service arm. If there’s an issue with Marauder or Corsair in the field, or some U.S. government customer is having a challenge with something, they’re the team on the ground that brings that information back in and closes the design loop with the engineering team.

In general, I think it’s working well. As it expands into Port Alpha in particular, we’re starting to brush up on some of these problems here in Franklin. But things like global supply chain become much more pressing issues.

For example, we wound up buying out a vast majority of the aluminum in the Louisiana area where we plan to build Marauder. I’ve been thinking about that critically and taking that lesson learned. I’ve reached out to critical suppliers for Port Alpha to make sure that the steel pipeline is in place, and I’ve been thinking about how we scale and move those rocks ahead of the avalanche that is about to come.

That’s a really good point. We’re on the precipice of reindustrialization.

I think we’re in it, yeah.

We’re literally directly in the eye of reindustrialization. Where do you see the biggest stress points on that, whether it’s supply chains, talent, or finding Port Alphas around the country?

3. Reindustrializing American Shipbuilding

There are a couple of key Lego blocks that need to fall into place here to make this successful. Dino has probably gone on about capital. That’s important, right? Part of this is a bit of a chicken-and-egg problem, where you’re not going to see movement in the market until you invest in the market and show those demand signals. We’re fortunate enough that we can do that a little bit ahead of need.

On the contracting side of the house, that’s something where I think we’re looking for support, love, and guidance from the U.S. government. The more that the government is willing to lean in, the more that commercial industry is going to follow suit.

On the executional side, I think it’s interesting. We have a philosophy in our production system that, if you think about the last 20 years in defense tech manufacturing, I won’t go as far as calling it gatekeeping, but there’s been a lot of focus around certifications and qualifications. What that has done is, candidly, create a quality system, and that’s important. It’s something that we want to retain. But it has kind of precluded a large portion of the labor force from reentering that market.

That’s a problem that should be solved systematically, whether it’s in how you deal with automated inspection, how you deal with work instructions, or how you generally acclimate somebody to a manufacturing environment. There’s no reason somebody who can change a tire shouldn’t be able to go and build a Corsair, work on a Marauder, or work on a larger ship. Break, break.

Labor and workforce are certainly a challenge. I think there’s a path there. There are plenty of Americans who want to work and want those kinds of jobs. The thing that’s going to be a little bit challenging is that there are a couple of key industries that I think have generally migrated outside of the United States.

Casting and forging is a big one. I see that being a challenge here in the next 3 to 5 years. The same thing applies to a lot of wire harnessing. Actually, we saw recently that a lot of harnessing was previously done in Ukraine or Mexico. We’re starting to see that come back in. All of these are things that need a good center of gravity in the United States to come back into domestic production for this to be successful.

That’s a really good point. We were just at Sundara Systems with Jordan Black, the CEO there, and he was talking about how they’re bringing wire harnessing to America to automate it. They saw this problem at SpaceX. It’s crazy and very complex. It barely works, it’s always a problem, and it’s a huge headache for everyone.

His point was more on the talent side of things. They created a certified training program there. It would usually take you about 2 years to learn how to make—

Yeah.

—complex wire harnesses, and they brought that training program down to about 4 weeks. So how are you, as you continue to scale out—

Yeah.

—I’m sure you’re going to be hiring tons of people. I don’t know if you guys can see it, but we’re in this very large warehouse that also doubles as your lunch spot.

Yeah.

You’re full of people. It’s men, it’s women, it’s young people—

Yeah.

—it’s older people. It’s all this mix. How are you bringing them in, giving them the skills, and giving them the confidence to build?

It’s interesting. If you had an opportunity to tour the floor, there’s actually an enormous wire-harnessing shop in the background. I know Jordan, and I know Senra. I wish them the best of luck. I think they’ll help solve that problem, but it is a problem set that’s larger than just one organization.

A similar approach applies to learning and development. The barrier to quality, I think, is the challenge that we all want to tackle: how do you lower the bar of excellence such that somebody with less experience, less talent, or less niche expertise can actually go and meet and exceed it? You can do that through better design and engineering. Candidly, I think that’s a place that everybody should be focusing on. Less parts are good parts. That’s certainly our mantra here at Saronic.

The other thing is the basics, right? You can do less upfront if your quality and inspection and your pass-through rate are much better on the back end. You can focus on automated inspection, digital odometry, and all those different things to make sure that the end process is good.

The last part of this is that you actually do just have to teach people. The way we think about it is a combination of traditional learning, so we don’t necessarily do classroom learning.

Adults candidly learn very differently from school-aged children. It’s more about risk and reward and teaching skills and outcomes. Doing that in partnership with the traditional apprenticeship kind of model actually works really well, as long as you can scale that up and make those connections quickly.

So, if we look to the future, what are you most looking forward to in the next 12 months? I know that—I don’t know if that’s going to be a long amount of time or a short amount of time for you—but you guys are growing really fast.

Saronic operates in dog years. That is the mental equivalent of 7 years in my mind. Gosh, what does the next year look like for us? That looks like us through construction in Port Alpha, which I’m very excited for. It looks like us standing up auxiliary facilities out in Brownsville, so think of more testing locations, more inventory, all that jazz.

Between you and me, I think we have a couple of products up our sleeve that we’ll be introducing in that timeframe as well. So—

Ooh.

—always excited for that. I think, more importantly, my hope is that within that year window, we can actually push the regulatory envelope, so you start seeing autonomous surface vehicles in your day-to-day life, much like you see Teslas and Waymos and self-driving cars on the streets.

Are you going to take one for yourself?

I’ve been in maritime long enough to know that owning a boat is never a good idea. Candidly, there’s not enough money in the world that I think Dino could pay me to own a boat at this point in my life. I’m good. I don’t need to take one.

Oh, my God. Okay, fine. Keep it to your customers.

Yeah. They’re more than welcome to do it.

So, as we have Dino, Rob, and Vib come on, what questions should we be asking them?

I mean, Dino obviously is the soul of the company, so certainly poke at that. I think you’ll get a phenomenal conversation with Rob about acquisition reform, so that’s something he’s very passionate about and certainly focused on. Vib is, candidly, probably 10 to 15 IQ points above me, so feel free to just talk to him about everything from Nietzsche to software to whatever you want to know about AI in the future. The gentleman can certainly speak to it.

His philosophy around software-hardware co-design, I think, is powerful and unique, and will give you insight into how these things can work in the future.

Doug, what is your hottest take right now?

Yeah, Molly, my hottest take is that I really do think we’re reentering the world of the here and now. What I mean by that is, I think being successful in real space, doing hard tech, building real things, and manufacturing with real tangible items is going to be the moat of the future, especially as software becomes a bit more commoditized and we see a resurgence in AI and bot coding.

Good hot take.

Good hot take.

Well, Doug, thank you so much.

You’re very welcome. Happy to be here.

Rob, welcome to Sourcery.

Thanks. Glad to be here.

Okay.

Looking forward to it.

So, we’re on the big tour of Saronic. This is a part of our series, and it’s been super fun. We’re talking to everyone from Dino—we just had Doug on—and we’re having you on now. You’re the chief commercial officer. We’re going to have Vib on next. But for those who aren’t aware of you and your background, you have a really crazy background. Can you explain a little bit of what you were doing prior to Saronic and how you got involved?

Sure. I’ve been a generalist ever since I was a kid, still trying to figure out what I want to do when I grow up. But when I finished Penn State, I knew I wanted to join the Marine Corps, so I became an officer, deployed a couple of times, and then joined the Reserves and wound up staying in for 23 years total.

When I went in, I figured I was doing 4 years and out to solve that—to figure out my life problem. I was actually completely out of the Marine Corps for about 9 months, and I really missed it. I missed the people, I missed the mission, so I went back into the Reserves. While I was doing that, I was working at a couple of defense primes. I took over a struggling company and then started my own consulting firm.

I was able to be on both sides of the line, where I could put the MARPAT cammies on and see it from the customer perspective, and then from industry, look at it through that lens and see the problems from a different perspective. It was a unique but painful experience that allowed me to make it here.

And so, how did that bring you to Saronic? How did that evolve into this role?

It was a pretty serendipitous moment, and there’s been a lot of serendipity involved with Saronic, where stars have aligned at the right times and right places. I was actually doing some consulting work for 8VC and Joe Lonsdale, and I had also just met Dino on a hunting trip. We were connected through a mutual friend, and we just hit it off, kind of veteran-to-veteran, amongst this group of people.

He broached the idea to me. I was helping him engage with parts of the Marine Corps to get some validation for the thesis, and then the famous line—I’ve said it a couple of times—is he called me and said, “Dude, I need a you. Do you want it to just be you?” And—

Aw.

Yeah, that was kind of the—

That’s so sweet.

—the beginning of the journey, and it has been a wild ride ever since. I picked up my family from Northern Virginia, where both my wife and I were born and raised, with very D.C.-centric family and background, and just uprooted to Austin. We didn’t know anybody except Dino, and, yeah, the rest is history. It’s been a wild ride.

Damn. So, what does day-to-day look like for you?

Quite a few airplane trips back to D.C., of all places. I have all of our growth functions, our program management, government relations, and then the functions that support those activities: contracts, proposals, and our international portfolio. So, the hair is rapidly graying.

It’s a tight flywheel, though. I think usually you separate the program operations function from the growth and business development group. But, A, we’re centralized in maritime. B, the autonomous maritime space is a pretty small community of interest, relatively speaking, so we made an intentional decision to have a really tight flywheel where my program teams are talking to customers. Oftentimes, they’re the exact same people our growth folks need to talk to.

It was a good way to deconflict fires and make sure that when we do get a feedback loop from a customer, it’s fed into our systems as quickly as possible to sort out a solution.

So, Saronic, I believe, closed government contracts within the first 90 days of founding?

One, and it was a cooperative research and development agreement.

Mm-hmm.

CRADA—a government term of art. It’s essentially a no-cost contract where, in exchange for codifying the relationship, you can get access to certain customer requirements or information that may not necessarily be public, but is not classified.

It’s basically a partnership where, in exchange for the government providing you insights, you’re acknowledging that they’re doing so, and anything that originated out of that, you have to manage according to the intellectual property guidelines. But at the end of the day, it was a peek behind the curtain that allows you to jump-start your process.

We did that the same way we do everything, which is just a ruthless seeking of insights from the fleet level. We didn’t approach our initial customers at the PEO or PAE level. We didn’t go from the Hill. It was really finding operators and understanding: What are the things you’re being asked to do that you have to say, “Yes, sir,” or “Yes, ma’am,” and then you look at each other and say, “We don’t have what we need to be able to do that”?

That is the gold standard for us, because we can take that, determine the best way to provide a capability that will address that need, and then we get to iterate with the customer and understand: Is this what you need? Do we need to make rudder adjustments? In some cases, you don’t get that feedback.

You get one piece of insight, and you've got to make a gut call. I think that's one thing working with Dino, Doug, and Vib that is great: four founders with totally unique backgrounds, and the Venn diagram overlaps but just a little bit. So we can make those gut calls, and we'll sit in a room, we'll slap the table, and then just ruthlessly execute.

A commonality between most of them so far—I don't know, maybe Vibhav will say the same thing—is that you're a huge fan of acquisition reform.

4. Reforming Defense Acquisition

I have scar tissue over a long and circuitous career path, so as I said at the beginning, I've seen it from both sides. When I was still in the Marine Corps, it was a great disarming mechanism where you could look at a contracting officer or a program manager and say, “Look, dude, I've been in your shoes. I understand the perspective you're looking at it from. Neither of us are boogeymen on either side of this line.”

At the end of the day, the real magic happens, acquisition-wise, when there's trust between both entities and a shared commitment to an end result. So there's a lot going on now. I've never been more excited about real change. The administration's done a fantastic job of pushing real reforms, putting real money behind them, and putting real teeth and consequences behind not adhering to them.

One of the things we've seen in the past in industry is an NDAA piece of language that ultimately yields a report that no one reads, and nothing's done about it, and people keep moving up the ladder. The thing we're seeing right now is there's visibility into where the breaks are in the system, down to an individual level.

If there's a contracting officer who's blocking something that a program manager, leadership, or a combatant commander are all aligned on, at the end of the day, you can't have one human who's not tied to the mission making arbitrary decisions based on a guidebook that doesn't have any wiggle room for them. So I'm a big fan of command relationships, and a program manager—or now a PAE—who has the risk levied upon them.

You are responsible for everything that does or does not happen in your unit. They teach us that the first days in the Marine Corps. How can that be true if someone on your team can overrule your decision and put their foot in the ground and say, “I'm not moving until this happens”?

We've been in those positions. Sometimes we've had to escalate it. Other times, we find a way to work around it. But I think at the end of the day, accountability is key.

Government employees—you can only incentivize them so much. We talk about aligning incentives. You can only punish them so much, too. There are really limited options there. The one thing you can do is put people in positions of accountability and then hold them accountable, because nothing's going to motivate them to get someone out of the way, modify a process or a person, and get things done.

The big one for me is, how can we take AI and some of these regulatory processes—agreements officers, contracting officers, and the acquisition professional community, mostly great people who are charged with adhering to a myriad of policies and regulations—and help them? When they're told, “You've got to do it,” I feel for them, because they have to abide by this rulebook.

I think one thing that could help is using AI to augment the acquisition workforce, because we've come across multiple instances where there's a lack of awareness about a new regulation or a new policy that tells them they can do something, but they are adamant that they can't.

How do you even fix that?

It's hard. You can't personalize it. When I started my career, I used to get angry at individuals, and I used to get angry at policymakers, but at the end of the day, you have to understand what someone's trying to do.

Sometimes a commander in the field can get stuff done when someone in the shore establishment can't, and vice versa. I think the magic happens when you connect the warfighter and the leader of a program—the leader of a portfolio. That's the quickest way to get everybody on both of their teams singing to the same tune and playing on the same page of music.

The music book's changing by the minute, so I feel for everyone. That's where I think having some tool for a contracting officer to say, “Yes, this is okay” could help, because right now I think they're falling back on what's built their careers to this point, which is playing it safe and adhering to old rules. In some cases, they're not even fully aware of the maneuver space they have to get things done quickly.

Wow, okay. So as we close out, what are you most looking forward to in the next 12 months?

Well—

Big sigh.

Yeah. I would say sleep. This has been a wild 4-year ride.

I think what we've seen in the last couple of weeks is tangible evidence that the technology we're creating is having an impact on real people and real operations. You can touch, feel, and see what that impact is. I think that was great for our workforce: to put something tangible in front of them, so they can see that the hours they're putting in are making a difference.

I'm really excited to see, as we move into shipbuilding, the same technological elements that are in the platforms behind us being put into larger and larger ships, because, frankly, that gets more and more people out of harm's way.

But I think what I'm looking forward to most is shaking the hands of those 2 Apache pilots. We still haven't found out who they are. Whenever current operations conclude, or we get a chance, that's one thing I've told Dino: We have to find a way to get connected to those guys, cameras or not, shake their hands, and thank them for what they've done.

Incredible story. It really was.

Yeah. It's an honor to be here, and it's great to take people who've never been in defense before and show them that there's a pathway to make a difference in the world.

Amazing. Well, Rob, thank you so much.

Thank you.

Vibhav, welcome to the show. We're almost 4 for 4 now.

Yeah. Thanks for having me.

For people who don't know you, you were the CTO of Saronic. How did you get here? What was the story and the background?

I grew up in Silicon Valley, in Cupertino. Naturally, everybody around me was—and my parents were also in tech—a lot of people around me were going into software. So naturally, when I went to college, I was like, “How do I choose not software to go and work in?”

I started off as a chemical engineer, made a brief detour and played cricket for a little bit, and then I came back to study electrical engineering.

Cricket?

Yeah, I played cricket for a while.

After college, I found myself at 8VC, a venture fund in San Francisco, which I guess is not too uncommon for electrical engineers to go into finance, but maybe—I don't know. I was at 8VC, working on a company there as part of their build program in the financial-services space.

I think I was there when we'd made the investment in Anduril. I'd interviewed at Anduril and ended up joining in late 2018. This was when the company had maybe 3 products, but the flagship one was the Century Tower. That's what I started working on.

I was there for around 4 years. I worked on a bunch of different things while I was there and had a great time. Then, in the summer of 2022, my old friends at 8VC had put me in touch with Dino. I ended up connecting with him and flew down to Austin a couple of times.

He had this grand vision of building autonomous boats. Back then, it was actually hydrofoils, but scaling into boats. I was like, “Sounds awesome. Let's go do it.” So at the end of 2022, early 2023, I ended up moving to Austin, and I've been here since.

I have to bring this up, but you said you worked at Juicero?

I did. I interned at Juicero in the summer of 2017. It was about 5 or 6 weeks into my internship that the company ended up shutting down, but I met some great people there, some of whom now work at Saronic.

Being at Anduril and being here, we kind of got a chance—the luxury, and we're grateful—to experience hypergrowth. Earlier in my career, I also saw the other side of how startups can go. But it was still a fruitful experience.

I was part of the Kleiner Perkins Fellows Program back then, and that's how I'd gotten introduced to the company. KP is back, and we have a good relationship with them. They're one of our big investors, so it's kind of a full-circle story.

So in just 4 years, you guys have scaled to a $9.25 billion valuation. How do you manage this day to day? What are you focusing on?

5. Scaling Autonomous Hardware

We don't necessarily wake up thinking about the valuation every day. The general focus is: How do we translate early, large capital investments into making a meaningful dent in economic capacity, production capacity, and the technology of the future?

For us, building ships—building hardware in general—is CapEx-intensive, but building ships in particular is extremely expensive.

And so really, the focus for us is: How do we translate that into understanding requirements and understanding what the technical needs are, and then layer that into what parts of the development cycle, design cycle, or execution need to be innovated on? How do we translate that into better economics?

What is your process for development here?

Yeah, it’s grown and evolved over time. Our first autonomous boat was a plastic box we bought from Amazon and put some electronics in. From there, we had a 6-foot autonomous boat, and we evolved that into what you see behind me right now, Corsair. All of the processes have changed over time, and I think a core element is that the software development life cycle for robotics has changed a little bit in that time.

In the past, when we were part of different embedded systems projects, your software development cycle was actually quite long, and you spent so much of your time debugging hardware. But you also had to spend the same amount of time writing code. Today, with AI tooling, you can write code a lot faster, but that also means your debug cycles are a lot faster. And so you spend a lot more time up front now—or it’s prudent to spend a lot more time up front—assembling all your hardware on a bench, testing it there, and investing more heavily in observability and simulation.

A lot of those things that you typically think about later in the hardware design cycle, you can actually start to invest in up front. For Marauder, for example, our 180-foot ship, we built all of the internal genset, propulsion stack, bow thruster, and all of those components of the ship in a room, probably 8 months before the ship was built. And so we could write a ton of that software. That let us reduce a lot of the iteration cycles around the delta between what works in simulation and what works in the real world. You’re just doing a lot more homework.

By the time the boat actually splashes, you’re directly into the application layer. You’re thinking about what the user experience is going to be and how to improve models—things that usually, at least historically, in a lot of the hardware products that I’ve worked on in the past, just didn’t happen as quickly. That process has evolved a lot.

One piece of advice I got earlier on was that software typically moves really quickly. Hardware engineers tend to measure twice and cut once; software engineers just cut until it looks right. So how do you slow down the pace of software so you’re a little more methodical? You think through a lot of the infrastructure choices up front, early on, and make very opinionated choices.

On the hardware side, how do you speed up hardware life cycles? That means you buy parts early, build hardware-in-the-loop systems, and try to reduce the number of requirements so that you can go faster. The nice, almost magical part about building autonomous boats is that all boats historically have been built for humans. When you thoughtfully choose to design hardware and software together, you actually get to reduce a lot of the complexity. So you can increase throughput and reduce the time it takes to actually build the system. There are a lot of different parallels that help us go faster.

So are you saying that you are helping with placement—figuring out where the best sensors will go? How do you co-design with the boat?

There are a couple of different areas. First and foremost is the electronics. The compute, the electronics architecture—a lot of the focus on the products that we build is around what the government refers to as modular open systems. Really, what that boils down to in a couple of different places is: Do you have easy-to-use software APIs and easy-to-use hardware APIs? Are these well-known protocols and well-known interfaces? If you have some arbitrary, weird-looking plug that somebody has to custom-design for you, it’s going to be very difficult to integrate arbitrary payloads and hardware.

Similarly, from a software standpoint, if you have some proprietary, bespoke protocol, it’s going to take someone forever to integrate. But if you have plug-in, more plug-and-play tools, it’s a lot faster to integrate payloads. We start at the electronics layer, making sure we’re choosing the right compute and the right interfaces, and making sure that the electronics are designed for redundancy in certain areas.

For example, a bunch of different cameras might be on a particular power distribution module, and that allows you to cycle the power to each and every one of them on and off. The luxury that gives you is that now, if you need to—engineering 101, right? Something’s not working, what do you do? You turn it off and turn it back on—you now have the ability to do that with software, remotely. These are just intentional design choices up front.

Similarly, for our products, you’re in the middle of contested RF environments where you’re deployed. A lot of times, you don’t want to turn on your radar because that reveals your location. Other times, you need to power-cycle for a variety of reasons. Maybe it’s a particular posture that someone wants to project, or something’s not working as you expected it to. There are a bunch of different electronics design decisions up front that help with that.

Then the next step is your sensor and perception stack. You probably don’t want to use USB cameras; there are a bunch of issues with those. How do you design the calibration mechanism for your sensors to all be well-calibrated with each other, so that when you actually get pushed out into the middle of the ocean and something moves around—because the ocean’s a tumultuous place—you actually have visibility and observability over that?

I would say the next phase of the co-design looks like what you said: sensor placement. Sensor placement can be cameras and radars, but for more complex systems, it could be where you put different RF antennas that are speaking at different frequencies so you don’t have electromagnetic interference between those systems.

It cascades down. You’re working backward from: What is a user? What are the core user stories or product outcomes of what you’re trying to build? Then you work backward from there, and there’s a natural overlap into the hardware design. Finally, from there, it goes down to the naval architecture: Where do you place your electronics cabinet? How many G-forces is that going to take in sea state 5 or 6? What are the downside protections you have against that? All of those things end up being part and parcel of the same core technology or product philosophy: How are we going to design this in a way that achieves mission outcomes?

What has been the biggest technical challenge?

6. The Hard Problems At Sea

There are numerous. I would say that the maritime industry, generally speaking, for the last several decades, has had components that are very, very analog. A lot of the systems have primarily been designed and mass-produced for humans, so they don’t have modern software APIs. You don’t have good simulation tooling.

I remember in the very early days, we reached out to a really big marine design firm in Europe. They were huge, one of the biggest ones, and we asked them for simulation software. We wanted to get a better understanding of how their software worked, to understand where we should really be starting. Every single time we tried to design a boat under 50 feet in their simulation software, there was numerical instability.

A lot of the hardware components as well as software components are just legacy, so naturally there are a lot of things you have to do from scratch in certain areas. The other big pieces that are just hard are building and fielding hardware quickly; it’s a difficult, arduous, painful process. You learn a lot when you put things in the ocean.

The ocean is also just a tumultuous environment. Saltwater hits your antennas, and you lose quality of service. You send a boat on a mission that’s 30 days long, and birds start nesting on the boat. You have biological growth that happens—there’s biofouling. There are all these different challenges involved in building and designing for the ocean.

I would say the other piece is that, when you think about UAVs and ground vehicles today, there actually is a large volume of open-source data and information, right? I think there’s a company that just published a massive open-source data set for perception in Ukraine. We’ve had research labs, and we’ve had the self-driving car industry. We’ve poured hundreds of billions of dollars into self-driving cars, and we’ve open-sourced a lot of great data, a lot of great research.

Maritime is very sparse. I think it’s for 2 reasons. 1, from a commercial standpoint, autonomous boats haven’t been blessed with the legal approval necessary from DHS and whatnot. You still have manning requirements on autonomous boats, so you haven’t had the economic push. On the other side, a lot of the things—as soon as you get into acoustic perception and things subsea—they just get classified really quickly.

Mm.

You just don’t have that open-source platform to jump off of. Whereas with UAVs, for example, you can go home and build a drone this weekend, and the parts and software are pretty commoditized, and you can get to something pretty good pretty quickly.

You’re not going to add lasers to the boat to fend off birds yet?

I think we might violate some laws by shooting birds with lasers. But it’s not like we’re going to do that.

The middle of the ocean is a crazy place. Being at the surface of the water in particular is a little bit more challenging because you’re in a dual-fluid environment, right? The surface of the ocean is really the tumultuous part. If you’re 10 feet below the surface, you’re in a single fluid; it’s fine. You can plan accordingly.

But at the surface of the water, you’re really thinking about that 1% of times where you catch a bad wave, or you go at the wrong speed in the wrong direction and stuff the boat or something. There are all of these niche, small-end-of-the-distribution problems. Your perception, the software, and all the hardware components have to be ready to hit that level.

A good example of this is that I think a lot of commercial boats target human beings to take maybe 0.4 to 0.6 Gs. Our boats, we’ve seen accelerations north of 20 Gs before.

What?

A human would not take that. This boat accelerates pretty quickly, too. It’d be very dangerous for a human to be on that if it’s going that fast.

Again, these are all design problems and challenges, and this is a newer field to really tackle them in. Whereas with drones, they’re always meant to have no humans, right? So designing for that from the beginning is different. For us, it’s slightly different because there’s this merge of some stuff taken from the manned world and unmanned pieces coming in. It can be quite tricky sometimes.

You have some exciting software-to-hardware partnerships. You have one with Palantir, Path Robotics, and NVIDIA. How did you evolve these, and what’s going on with NVIDIA?

With NVIDIA in particular? Yeah. Most of the robotics compute today probably runs on some flavor of GPU compute, and NVIDIA dominates that market.

NVIDIA has been a really powerful partner for us, not just in terms of procurement and helping us make good decisions when it comes to a lot of the open-source tooling, but also in some friendly conversations around what the next-generation simulation tooling will look like. The ocean has just been hard to model forever. The weather is hard to model. These have just been active research problems for a long time. But they’ve been a great partner there.

Then, again, from a compute standpoint, NVIDIA is the one we go with, and I think that’s what a large number of the robotics companies go with, too. I think it’s a combination of price point, but predominantly, the software layer is just the easiest to move quickly with. Because they have a large market share on the training side, obviously the inference side becomes easier to seamlessly transition into. So that’s the nature of our partnership there.

With Palantir, it’s predominantly related to internal business systems with Foundry, to accelerate manufacturing and manufacturing effectiveness. With Path Robotics, it’s partnerships down in our shipyard at Gulf Craft in Louisiana, to help with accelerating welding and some of the workflows there.

You mentioned redundancies. I can imagine communication bandwidth is pretty hard on the seas. Do you do anything with Starlink? How do you manage those?

Yeah, Starlink, and then Starshield is the defense flavor of it. All of the platforms that we build have what we call a PACE plan. It’s basically a redundant set of communications, so that if something goes wrong with one of those systems, you always have a way to communicate. There are a variety of both beyond-line-of-sight and line-of-sight RF payloads or devices that you can use to do this.

But the main reason for this is that, on one side, if you think about this as a drone, you have active jamming. You can have all of these contested RF environments. The nice part about the ocean—or maybe the bad part about the ocean—is that the ocean kind of does that for you regardless.

If you’re in a heavy sea state, and water is falling on your antennas and whatnot, you’ll already see that. The ocean’s trying to jam you, basically. You’ll see a quality-of-service drop. You’ll see dropped packets and things like that. So it’s really important for us to have that redundancy; otherwise, if there’s a human in the loop or a human on the loop, you want to have some observability. You need the ability to see what’s going on.

The last piece is that if part of the mission is, “I need to go to this location and then turn on all of my comms,” you need the ability to do that as well.

So if everything goes out, will the boat be able to find its way back? Will it complete the mission?

Yeah, it’s a good question. First and foremost, that decision is up to the operator in terms of how they want to design the mission. You can use our platform with the open APIs that we have to design a mission according to your specifications.

You can say, “If I lose comms, I want to keep going.” Great. “If I lose comms for X amount of time, then I want to actually turn around and come back, or I want to go back to the last location where I had comms.” You can mission-plan whatever arbitrary logic you want.

Ultimately, that responsibility is up to the operator. They can choose how they want to orchestrate that. In certain cases, operators might be like, “I know I’m going to lose comms.”

I want you to continue. If I get comms back, I get them back, and if I don’t, these are the stopgaps that I have in place. Other times they’re like, “Hey, this is internal testing. We actually want to make sure we know where the boat is all the time. If the boat loses comms for X amount of time, I want it to loiter or return to base, or whatever it might be.”

Cool. Okay, as we close out, two more questions. What are you most looking forward to in the next 12 months? Doug said that’s a little bit ridiculous. You guys operate in dog years.

Yeah. I think there’s a big difference in the operational cadence when you have hundreds of something versus when you have, like, 5 or 10. When there’s this amount of mass proliferation that we can build out there—when we build 20 Marauders or hundreds of Corsairs—the volume actually changes the core fundamentals of how that system gets used.

If you only have 1, you’re babying it the whole time, but when you have hundreds, there are all sorts of interesting things that you can do. I’m really excited about that. Obviously, Marauder is a beast in and of itself. Seeing it on the water is just absolutely magical.

Getting to the point where you have a fleet of these that are completely unmanned has never been done before, so I’m really excited about that. The last big piece is that we’re continuing to scale the company and open up new markets, both in defense and commercial.

At a high level, the thing that I’m most excited about is that I still don’t feel like the aha moment has really clicked for a lot of the different markets: if you reduce the overall cost of doing things in the ocean, there are so many new, cool things that you can get into. Right now, there are just a lot of places in the ocean where it’s prohibitively expensive.

I’m really excited about some of the new markets over the next 12 months.

Cool. Hopefully you guys find some U-boats in the meantime.

Yeah, we’ll see.

Okay, so this will be the most difficult question.

Yeah.

What is your hottest take right now?

It’s not a hot take, I think, but I’m pretty excited about the wave of open-source AI.

I was going to ask you where you stand on this.

Yeah, I’m pretty excited about that.

Okay.

I think one of the reasons why technology is at the place that it is today for us—whether it’s your Oura Ring, your iPhone, or anything—is owed to open-source technology. It’s surprising how much of the internet is based on a handful of open-source repositories that great engineers have created, not for seeking a bunch of money on the other side, but really just for pushing the needle of technology forward.

I’m excited about the wave of open-source AI. I think it’ll democratize a lot of the tooling and resources. I think we’ll continue to see proliferation on the hardware side and the capital-expenditure build-outs there.

The exciting part, I think, to everyone is that those are the kinds of competitive environments that make technology better for everybody. I’m excited about that. It’s scary, but it’s cool.

It’s scary?

I mean, it’s cool.

It’s—

It’s not just scary. It’s cool.

Okay. Well, Vib, thank you so much. This was awesome.

Cool. Thank you.