How US Hardware Startups Will Outcompete China | Pari Singh, Flow

Summary

Pari Singh, founder and CEO of Flow Engineering, shares his journey from building rocket engines to creating software that is transforming how hardware companies design complex systems. Starting as “The Rocket Company,” Flow discovered that the fastest design consultancy could go from requirements to detail design in 12 weeks - but they could do it in 2.5 hours through their integrated platform. This led to the realization that hardware engineering needs a fundamental paradigm shift from waterfall to agile, iterative approaches.

Singh makes a bold prediction: there’s an 85% chance of US-China conflict within 5 years, and the US can only win through superior reliability/safety and AI/autonomy capabilities. He argues that traditional defense primes like Boeing and Lockheed are fundamentally unable to adapt because their waterfall methodologies cannot handle modern systems complexity. Companies like SpaceX, Anduril, and the “El Segundo” wave represent a completely different approach - one that operates more like software companies than traditional hardware companies.

The conversation explores three eras of manufacturing (analog, digital, iterative/autonomous) and how Flow is positioning itself to accelerate the entire hardware industry’s transition. Singh shares philosophical insights on hiring “second order” people who take local hits for higher slopes, and why hunger matters more than IQ.

Highlights

”The fastest design consultancy could do it in 12 weeks. We could do it in 2.5 hours.”

yt-dlp --download-sections "*1:52-2:55" "https://www.youtube.com/watch?v=989_qK5dVTU" --force-keyframes-at-cuts --merge-output-format mp4 -o "989_qK5dVTU-1m52s.mp4"

“The fastest design consultancy in the world for hybrid rocket engines could go from requirements to detail design in about 12 weeks. We could do it in 2 and a half hours. And the reason is we built this internal platform that integrated requirements to Matlab, Matlab to CAD, CAD to CFD and FEA.” — Pari Singh, 1:52

”China’s Flow equivalent is just hardcore mode with a gun to your head”

yt-dlp --download-sections "*3:52-4:55" "https://www.youtube.com/watch?v=989_qK5dVTU" --force-keyframes-at-cuts --merge-output-format mp4 -o "989_qK5dVTU-3m52s.mp4"

“China’s Flow equivalent is just hardcore mode. 996 is easy mode for China. Hard mode for China is significantly more with a gun to your head and a gun to your family. And the cost of inadequacy there is very very high.” — Pari Singh, 3:52

”I think that entire generation of traditional primes will die out”

yt-dlp --download-sections "*6:12-7:15" "https://www.youtube.com/watch?v=989_qK5dVTU" --force-keyframes-at-cuts --merge-output-format mp4 -o "989_qK5dVTU-6m12s.mp4"

“Over the next 10 years I’m actually very bearish on the traditional primes. I basically short all of those. I think that entire generation will die out. And this new wave of El Segundo companies will be successful because of the way they work rather than what they’re bringing to market.” — Pari Singh, 6:12

”Starliner was more expensive, took longer, AND was lower quality”

yt-dlp --download-sections "*9:03-10:05" "https://www.youtube.com/watch?v=989_qK5dVTU" --force-keyframes-at-cuts --merge-output-format mp4 -o "989_qK5dVTU-9m03s.mp4"

“Boeing Starliner was more expensive, it took significantly longer, but the most interesting thing is it was actually lower quality. And the reason is they tried to take this old school waterfall model on something that is so high complexity it requires a new way of working.” — Pari Singh, 9:03

”There’s an 85% chance that the US and China go to war in the next 5 years”

yt-dlp --download-sections "*15:07-16:10" "https://www.youtube.com/watch?v=989_qK5dVTU" --force-keyframes-at-cuts --merge-output-format mp4 -o "989_qK5dVTU-15m07s.mp4"

“I think there’s an 85% chance that the US and China go to war in the next 5 years. I think it’s like that high. Xi Jinping has been incredibly explicit around it. There’s also a great book by Ray Dalio called The New World Order about the rise and fall of superpowers.” — Pari Singh, 15:07

”If I could trade 20 IQ points for 20 hunger points, I’d make that trade”

yt-dlp --download-sections "*49:35-50:40" "https://www.youtube.com/watch?v=989_qK5dVTU" --force-keyframes-at-cuts --merge-output-format mp4 -o "989_qK5dVTU-49m35s.mp4"

“If I could trade myself 20 IQ points for 20 hunger points, I’d make that trade. Smarter people see more ways things can go wrong. They often don’t see what could go right.” — Pari Singh, 49:35

Key Points

• Engineering is Broken (0:14) - The way engineering is done hasn’t fundamentally changed since the space race; the biggest problem isn’t designing rockets or nuclear reactors, it’s how companies work
• 12 Weeks to 2.5 Hours (1:59) - The fastest design consultancy could go from requirements to detail design in 12 weeks; Flow did it in 2.5 hours by integrating requirements, Matlab, CAD, CFD, and FEA
• Abstraction Not Automation (2:21) - They thought they built an automation box but actually built abstraction - a single way of designing hardware that enables software-style agile approach
• China’s Flow is Hardcore Mode (3:54) - China doesn’t have Flow equivalent tools; they compete through 996 work culture and extreme intensity with “a gun to your head and your family”
• Waterfall vs Agile (4:52) - Generational shift from NASA/Boeing 10-year program cycles with Gantt charts to SpaceX/Anduril bottom-up 3-month cycles with aggressive design-build-test-fly
• Traditional Primes Will Die (5:47) - Very bearish on Boeing, Lockheed; predicts the entire generation will die out because of how they work, not what they build
• Complexity Breaks Waterfall (6:56) - When unknown unknowns exceed known unknowns, top-down Gantt chart planning fails; you have to go bottoms up
• Apollo Was Actually Agile (10:08) - NASA didn’t write a million requirements first; they went Mercury, Gemini, Apollo - iteratively approaching the moon through 25+ iterations
• Starliner vs Dragon (8:43) - Boeing Starliner was more expensive, took longer, AND was lower quality because they tried waterfall on a system too complex for it
• 85% Chance of US-China War (14:35) - Singh predicts 85% probability in next 5 years based on Xi Jinping’s statements and Ray Dalio’s “New World Order” analysis
• China is Ahead (15:15) - China leads in shipping, nuclear/SMRs, solar, batteries, EVs, robotics, human robotics; US only leads in launch and autonomy
• Two Ways US Wins (17:02) - Reliability/safety (Chinese products may kill users) and AI/autonomy (the mesh layer where drones talk to robots talk to submarines)
• The New Game (19:25) - Human robots fighting with drones, autonomous submarines, autonomous tanks working as one AI-controlled cluster - that’s what wins
• Three Eras of Manufacturing (23:25) - Analog (Apollo), Digital (Space Shuttle), Iterative/Autonomous (Falcon 9) - each era requires reinventing how engineering is done
• Software Invading Hardware (24:05) - Watches became computers on wrists, cars became autonomous computers on wheels; software complexity now 10-100x greater than hardware
• Responsible Engineer (RE) (25:23) - Key El Segundo term: mechanical/domain engineer + systems engineer + project manager + manufacturing person all in one
• Requirements as Trojan Horse (32:43) - Flow discovered their revolutionary platform only succeeded when disguised as requirements management - solving today’s problem to earn trust for tomorrow
• Horseless Carriage Effect (31:13) - People don’t understand new things; first cars weren’t “cars” but “horseless carriages”; Flow had to frame their platform as “requirements management plus”
• Stoke Case Study (32:09) - Stoke Space (now raised hundreds of millions) was 5 people when they became Flow customer; got to first reusable rocket engine with just 10 people
• IBM is the King (41:05) - In systems engineering, IBM dominates with archaic software built in the 70s-80s; an entrenched oligopoly of Siemens, Dassault, Autodesk, PTC
• Special Forces Culture (43:36) - Army rank-and-file vs special forces elite require different tools and processes; Flow deliberately chose special forces culture
• Second Order People (44:40) - 80% are zeroth order (school-job-retire), 15% are first order (McKinsey highrises), 5% are second order exponentials who change the world
• Jama Has 3000 People (44:15) - Flow’s competitor Jama has 3000 employees; Flow’s entire org was 8 people at Series A, yet they win head-to-head every time
• Take Local Hit for Higher Slope (45:50) - Very few people will take worse salary, longer commute, more hours for higher growth slope 2 years later - but those are the people who change the world
• IQ is Overrated (49:35) - Would trade 20 IQ points for 20 hunger points; smarter people see more ways things can go wrong, often don’t see what could go right
• Action Produces Information (51:10) - Don’t spend years thinking about highest value activity; locally optimize, take data from world, easier to come out of local optimum than do nothing
• Chinese Farmer Parable (52:30) - Boy gets lost (bad), finds donkey (good), breaks leg (bad), doesn’t go to war (good) - the wise farmer says “we’ll see” to everything

Mentions

Companies

• Flow Engineering (0:05) - Singh’s company reinventing how humanity designs complex hardware
• SpaceX (4:37) - Example of new wave hardware company operating like a software company with agile methodology
• Boeing (5:42) - Example of traditional prime that Singh is very bearish on
• Lockheed (5:42) - Traditional defense prime that Singh predicts will die out
• Anduril (5:08) - New wave defense company; really a software company with hardware components
• Joby (5:08) - New wave hardware company mentioned as operating differently from traditional primes
• Figer (5:08) - New wave hardware company in the El Segundo wave
• Stoke Space (32:09) - Rocket company customer, was 5 people, now raised hundreds of millions
• McLaren (29:08) - European company Flow tried to sell to; overwhelmed and 3 months behind on current vehicle
• Rolls-Royce (29:08) - European company Flow approached; recognized Flow as next platform change after CAD
• NASA (28:10) - Tried to build similar technology for three decades; called it “intelligent census environment”
• IBM (41:05) - The monopoly king in systems engineering with archaic software
• Siemens (40:40) - One of the four big CAD vendors in the oligopoly
• Dassault (40:40) - One of the four big CAD vendors
• Autodesk (40:40) - One of the four big CAD vendors
• PTC (40:40) - One of the four big CAD vendors
• Jama (44:13) - Flow’s competitor with 3000 employees; Flow wins head-to-head every time
• BA Systems (0:25) - Where Singh worked and saw engineering was broken
• BP (0:28) - Where Singh worked early in career
• Relativity (3:32) - Space company mentioned alongside SpaceX as example of impact Flow could have
• Radiant Nuclear (37:08) - Example of deeply iterative company where requirements are fixed

Products & Technologies

• GitHub (12:25) - Referenced as analogy; “if you turn off GitHub today in software, the entire industry will crumble”
• CAD (2:15) - Computer-aided design; Flow integrates with it
• CFD (2:15) - Computational fluid dynamics; part of Flow’s integrated platform
• FEA (2:16) - Finite element analysis; integrated into Flow
• Matlab (2:13) - Engineering software integrated into Flow’s platform
• IBM Doors (10:30) - Requirements management tool NASA didn’t use for Apollo
• SMRs (15:25) - Small modular reactors; China is ahead
• Falcon 9 (24:55) - Defining product of the iterative/autonomous era
• Dragon (8:40) - SpaceX spacecraft that outperformed Boeing Starliner
• Starliner (8:35) - Boeing spacecraft that took longer, cost more, and was lower quality than Dragon

People

• Pari Singh (0:00) - Founder and CEO of Flow Engineering, mechanical engineer by training
• Elon Musk (12:00) - Referenced as one of greatest entrepreneurs; playbook for building incredible companies
• Xi Jinping (14:50) - Has been “incredibly explicit” about intentions regarding Taiwan
• Ray Dalio (14:58) - Author of “The New World Order” about rise and fall of superpowers
• Brian Armstrong (50:53) - Coinbase founder who said “action produces information”
• Paul Graham (49:55) - Referenced for insight on intelligence vs determination tradeoff
• Balaji (25:58) - Referenced for rat-in-prime-number-maze analogy on Lex Friedman podcast
• Usain Bolt (47:55) - Set 100m world record after eating 12 chicken McNuggets; talent plus training
• Mark Zuckerberg (47:02) - Referenced for preferring to be underestimated rather than overestimated

Concepts & Frameworks

• Responsible Engineer (RE) (25:23) - Key El Segundo concept combining multiple roles
• 996 (3:58) - Chinese work culture (9am-9pm, 6 days/week) - Singh says this is “easy mode” for China
• A2AD (0:52) - Anti-access area denial bubble China has built

Surprising Quotes

“I think there’s an 85% chance that the US and China go to war in the next 5 years.” — 14:35

“China’s flow equivalent is just hardcore mode. 996 is easy mode for China. Hard mode for China is significantly more with a gun to your head and a gun to your family.” — 3:54

“If I could trade myself 20 IQ points for 20 hunger points, I’d make that trade.” — 49:35

“The fastest design consultancy in the world for hybrid rocket engines could go from requirements to detail design in about 12 weeks. We could do it in 2 and a half hours.” — 1:59

“Our competitor Jama has 3000 employees. When we raised our series A, the entire organization was eight people and we were aggressively profitable. We win head-to-head on every deal we take together.” — 44:05

Transcript

0:00 This is Pari Singh and he is the founder and CEO of Flow Engineering. You started out building rocket engines and now you’re working on software. What went wrong?

0:14 It’s a great question. So as you mentioned, I’m an engineer. I’m a mechanical engineer. I became an engineer because I wanted to build machines that mattered to humanity. And I went into the industry and I went to companies like BA Systems and BP. And what became really apparent to me was the way that we do engineering is broken. Hasn’t fundamentally changed since the space race. And the single biggest problem in all of engineering isn’t the design of reusable rockets. It’s not the design of nuclear reactors. It’s not the design of drones. It’s the fundamental way that all of these companies are working.

0:49 And in every industry, you see the same transition in this pattern, which is there’s an old way of working. It works for a bit when complexity is low. Complexity goes up, it breaks. We need a new way of doing. And in every single one of those industries, there’s this fundamental paradigm shift that happens. And that paradigm shift in how these companies work is long overdue in our industry. It needs to happen. There’s no way you can do the next era without it. And nobody was doing it.

1:14 So when we started the company, we were called the rocket company. We knew nothing about tech, about VC. We weren’t trying to build a startup. What we said is look, the way that engineering is done is broken and we need a new way of doing engineering. And we think the world isn’t ready for it yet. So what we’re going to do is we’re going to build a little corner of the universe. We’re going to call it the rocket company and we’re just going to design rocket engines. And rather than trying to sell software, we’re just going to invent this new way of working. And we’re going to be so much more productive than anybody else in the industry that we’re going to have a huge advantage in cost and then we’re going to take that advantage and build a business around it.

1:52 But the real essence of it is we wanted to do engineering right in the way that we thought it could be done. And that’s kind of what happened. So we did it for a year. The fastest design consultancy in the world for hybrid rocket engines could go from requirements to detail design in about 12 weeks. We could do it in 2 and a half hours. And the reason we could do it in two and a half hours is we built this internal platform for ourselves called Flow that integrated our requirements to our Matlab, our Matlab to our CAD, our CAD to our CFD and our FEA, and then it fed it back into the 3D models in that lab. And what we thought we’d built was an automation box. But what we actually realized we built was abstraction.

2:26 So what we realized we’d built is a single way of designing hardware where we could start with really simple low fidelity models and then three, four, five times a day we could increase the complexity of it and we could bring a software style agile approach to designing complex hardware. And we had this huge light bulb moment and we sort of realized it together. And what we realized is two things. The first thing is there is no way you can design complex systems without working this way. It’s just like this is going to be the way that everyone designs it. It’s so obvious to us. And the second big realization we had is it was so painful to do that way that we actually had to invent a new layer of tools. And either every company in the world needs to go and invent these tools and work out this paradigm and that’s going to take them 10 years, or we can just put it in a box and give it to them.

3:15 So what Flow is today, our mission is to reinvent the way humanity designs the most important machines. We think of ourselves as an accelerant to the entire industry. If we’re successful, the entire industry goes from this to this. Like if I was the best engineer at Anduril or at SpaceX or at Relativity, I could affect the course of the company by about this much. At Flow we get to do this and we get to do it for the entire space industry, for the nuclear industry, for the robotics industry, for the EV industry. If the war with China happens, we get to play our part within that too. And it’s just the most inspiring thing I could be doing in my life.

3:52 Is there a Flow equivalent in China? China’s Flow equivalent is just hardcore mode. It’s just unbelievably ultra hardcore. Like just amazing entrepreneurs basically and teams. But it’s also like 996 is easy mode for China. What’s hard mode for China? Hard mode for China is significantly more with a gun to your head and a gun to your family. And the cost of inadequacy there is very very high. So they don’t do stuff right, but the consequences are so high and the intensity is so big that they can afford to do that.

4:28 Yeah. They’re pushing through even on slightly suboptimal tools and stuff. Totally. Do you want to talk about how SpaceX in the early days went through design and figuring out hardware and getting those first few rocket engines built?

4:45 Yeah. SpaceX is a really interesting story. And I think it’s a good backdrop to the fundamental change that’s happening in the industry. So I’ll tell you about the fundamental change, then I’ll tell you a couple of stuff I’ve read about SpaceX. So, there is this generational change happening in the hardware engineering manufacturing industry. And it’s a change from old school waterfall to new school agile iterative. Think NASA and Boeing with 10-year program cycles in big Gantt charts to SpaceX or Anduril which are not top down. It’s not 10 years. It’s bottoms up. It’s like 3 month cycles and it’s aggressively design build test fly integrate and you don’t do one big cycle. You do thousands of little cycles and you integrate aggressively again and again and again.

5:35 And I think a lot of people have spoken about why SpaceX has been successful and there’s like a couple of obvious reasons: Elon, vertical integration, the fundamental economic issues around traditional primes. But I think the thing that most people miss is the real thing that’s made SpaceX successful is their approach to developing rockets and developing hardware is completely different to the way that traditional primes develop hardware. SpaceX and Anduril and Joby and Figer and all these amazing new companies operate much more like software companies than they do like traditional hardware companies. And it’s in that new way of working that creates all the productivity and all the returns.

6:12 And I think over time you’ll see this play out. Like I think over the next 10 years I’m actually very bearish on the traditional primes. I basically short all of those. I think that entire generation will die out. And this new wave of El Segundo companies will be successful. And that is primarily because of the way they work rather than what they’re bringing to market.

6:35 I want to kind of spend a moment on why do you think the traditional primes, let’s say Boeing, Lockheed, all those guys are basically not going to be able to evolve into this next state?

6:47 Great question. There are socioeconomic factors and there’s technology factors. And I think people have spoken about the socioeconomic factors with cost plus and how these companies fundamentally approach risk. I think the thing that’s overlooked is actually systems complexity. Let me give you one example. The software engineering industry used to be waterfall. Think Windows 2000, Windows XP, pre-five-year development cycles, huge QA teams, big waterfall Gantt charts, you burn the whole thing down, you test it, and then you ship it. And actually what looks like hardware today. Exactly. That’s how hardware is done today. And that doesn’t work.

7:28 And the entire software engineering industry moved from waterfall to agile. And the reason they did that wasn’t to do with the fact that it was faster or cheaper or produced high quality software. It’s because when you’re designing complexity at a certain level, you can’t design it in a Gantt chart anymore.

7:47 So let me give you an example. If you’re designing a simple rocket or you’re designing like an ICBM or something relatively low complexity, you can sit down in the first two months, you can write down a list of requirements and you can decompose the requirements from the top level to the bottom level and you’ve done this before and you can kind of be order of magnitude approximately correct around all those assumptions and therefore you can make those assumptions to a Gantt chart and you can follow the Gantt chart and all your assumptions turn out to be right.

8:18 When you’re designing super high complexity systems, the unknown unknowns are way bigger than the list of known knowns or list of known unknowns. And therefore, when you do that decomposition and you go from top level all the way down to what does the engine need to produce, you don’t know yet. And you don’t know yet because you haven’t made it. So how could you know? And you haven’t run the experiments. And when you multiply that by the systems complexity, i.e. the engine affects the structures, affects avionics, affects the software, affects the structures, affects the engine. When you have that kind of complexity, everything needs to run in parallel. Everything’s going to be wrong all the time and you kind of need to make progress anyway.

9:03 So when you’re designing a sufficiently high complexity system, the top down control complexity, put it in a Gantt chart, doesn’t work. And you have to go bottoms up. There’s no fundamental way that you can’t go bottoms up.

9:18 So if you look at every major industry, the software engineering industry moved from waterfall to agile in the 2000s. The chip design industry moved from waterfall to agile. We went from 100 transistors in a chip to over a billion. The gaming industry, great example, went from waterfall to agile. Mario used to be in a big ad. You can’t design super complex video games anymore in that way. And now that generational change is happening in hardware.

9:48 So when you look at something like Boeing Starliner versus Dragon, Starliner was more expensive, it took significantly longer, but actually I think the most interesting thing on Starliner versus Dragon is actually it was lower quality. And the reason it was lower quality despite spending like 2x and 4 years longer on developing it is cuz they tried to take this old school waterfall model on something that is so high complexity that it requires a new way of working. So if you’re designing complex systems and you want them to be safe and you want them to be reliable, you have to go bottoms up. You can’t go top down.

10:28 Do you have any examples of projects or hardware projects basically in the past that could have happened much faster if they were using something like this today?

10:38 All of Europe. The counter is actually even more interesting. So the counter question is have there been examples in the past where you’ve had super high complexity systems and you’ve had to work in this way rather than waterfall? And people think old school waterfall, new school agile. Actually if you look at Apollo, Apollo is the greatest example I’ve ever got of agile. NASA didn’t sit down and say we’re going to put a man on the moon and the way we’re going to do that is we’re going to go to IBM Doors and we’re going to write a million requirements and then we’re going to work out what could go wrong with a million requirements and do a DFMEA and then write down another million requirements and then add some fudge factor and then spend 20 years designing Apollo. Like that’s fundamentally not how we did it.

11:22 The way we put man on the moon is we said great, we need to go bottoms up because we’ve never done this before. It’s too high complexity. We don’t know the unknowns. So the first thing we’re going to do is Mercury, and we’re going to print it up and we’re going to stick a man on the top of it and we’re going to see what happens. Good luck. And then the next thing we’re going to do is Gemini. And what Gemini is is going to understand multi-stage. And we’re going to put one rocket. We’re going to put another rocket on top of that rocket. Then we’re going to put a man on top of that. And we’re going to go into space. And then we’re going to go into orbit. And then we’re going to have Apollo. And Apollo is going to be three stages. And we’re going to be able to go into orbit. And we’re going to be able to do some rendezvous maneuvers. And then we’re going to be able to come down.

12:02 And it took sort of like an onion theory of risk. It said, what is the first thing that we can attack? And then what is the second thing? And then what is the third thing? And then we iteratively approached it. And we didn’t do it in one big iteration. We did it in like 25 different iterations. And every single iteration we did some stuff, we learned, we tested. Some things went wrong, some things went horribly wrong. But on every single one of those, we leveled up and eventually we got to the moon, but we did that in an iterative and an agile way.

12:35 What is going to be the biggest differentiator from teams of the past and how teams in the future move much much faster?

12:44 What I see in a lot of traditional primes is analysis paralysis. They start asking questions around what could go wrong and then need to build strategies on how to derisk those before we’ve even got a concept product out. What I see working amazingly well in El Segundo in this new wave of hardware companies is we don’t overthink. We go what can we produce in the next 3 months? And if we can’t produce something in the next 3 months, then we’re dead as a business and we’re not going to do anything. And then 3 months goes by and you produce something and sort of some things work, some things don’t work. And then you go, how can we improve it? And we take this fundamental bottoms up approach to designing. And we realize that learning can only happen iteratively by doing stuff, accepting probability of failure and being okay with that and building enough of a company and a mission where I have multiple shots on goal. It’s not just one shot on goal.

13:38 Do you think that it’s going to require people like Elon to tackle these massive problems like SpaceX in the future or is this type of iterative philosophy going to kind of help much more founders build massive SpaceX like companies?

13:55 I think there’s two elements to this. Element number one is how do you build a generational company and element number two is how do you design complex hardware? And they’re kind of two different things. So Elon is one of the greatest entrepreneurs of all time. I think he’ll continue to be that in the next couple of decades. And he sort of built this playbook for building incredible companies. I think that is actually a separate story to next-gen hardware.

14:18 In the same way today, if you’re building a B2B banking platform or you’re building a B2C dating app, you do it in an agile way. You use GitHub, you use modern tools, you ship an MVP, you put in front of users, you iterate it, and you make it better every single day, and you don’t overoptimize for scale when you’re early on. That same approach is going to become the de facto approach that all hardware is produced. Whether you’re building smartwatches or human robotics or nuclear reactors or reusable rockets, this fundamental new way of working will become the macro global way of working. And then in that you also need the recipe to build a generational company.

14:58 You briefly mentioned China before we started rolling and you said that you’d like to talk about that a little bit. What about that?

15:07 I think it’s the single most important thing that people in our industry don’t talk about. And there’s a bunch of rhetoric. There’s also a bunch of implicit stuff and I just don’t think the industry is being as honest as it could be around it. So I’ll start by saying I think there’s an 85% chance that the US and China go to war in the next 5 years. I think it’s like that high.

15:28 And what leads you to that? There is a bunch of stuff that Xi Jinping has been incredibly explicit around. There’s also a bunch of stuff on the US side where I’m not sure Trump wants another election anytime soon. I think if he had to make a tough decision between fighting an aggressor and that, it’s unclear which way he moves. There’s also a great book by Ray Dalio which is called The New World Order and it talks about the natural rise and the fall of different generational superpowers over time and very clearly we’re in the US superpower right now and China’s coming up.

16:00 If you look at our industry in particular, the manufacturing, the hardware industry, it’s really clear to me that China’s ahead. And it used to be a couple of years ago that they were up and coming and they were starting to do some things better than us. And now I’d argue the vast majority of sectors and verticals, whether it be shipping, nuclear, SMRs, solar power, battery technology, EVs, and they’re starting to do some stuff around autonomy. They haven’t quite caught us up in launch, so they’re not there yet. I don’t think they’re fully there on autonomy either. But I think in most of the key verticals, robotics, way further ahead than we are. Human robotics, way further ahead than we are. They are significantly further ahead and the pace that they are operating at candidly is faster than the pace we are operating at. China doesn’t use US chips. The US uses China’s chips. And China has more control over our supply chain than we sort of think is real. So, in a world where you hope for the best and you plan for the worst, I don’t think we’re quite being as intellectually honest as we can be right now.

17:05 How should we be operating?

17:07 So I think the right question to ask is in the, we should all avoid war and we should try and make sure that that never happens. If we hope for the best and plan for the worst, the question is if we go to war, how do we win? China’s ahead on most hardware stuff. In a world where it’s manufacturing, how does the US and how does the west win that war? And I think there are two key answers in which we can win. The first is reliability, safety, whatever we call that sector of stuff. And the second is autonomy and AI. Those are the two big things.

17:40 So let’s talk about reliability. China’s further ahead than we are. Would I and they’re going to be first to market for EVTOL. It’s a great example. Now would I feel safer in a Chinese EVTOL or an American EVTOL? Well, the US might be further behind, but I trust the reliability of the US and the FAA way more than I trust… way more like regulations and difficulty to get to market sort of thing. Yeah, there’s an onus on reliability and safety here and it’s not quite as conservative as Europe, but I’ve got feelings around that, but it’s also not quite as relaxed as China. And I think this is a really key area where if China’s first to market with a significantly inferior product and we’re slightly slower to market, but we win, we win what actually works, then there’s a real option for us to win.

18:32 Like Russia was the first to space. Russia beat the US with Sputnik, but the US got the first person on the moon. And I think there is something analogous to that in the sense that China will probably be the first to market on a number of key developments. But if their SMRs blow up and kill a bunch of people and there’s a huge amount of radioactive waste in the atmosphere, if the EVs are killing passengers on a regular basis, if the EVTOL are dropping out the sky, then we’re in a better position by being a couple years later to market but an order of magnitude more rigorous and more safe with our products.

19:08 My only worry with that would be that people, if you have the ability to gather all that data because you have stuff actually working, then you can probably iterate faster on making the product better. So even though there may be this lag where they start out worse, but are shipping faster, very rapidly, kind of like SpaceX, they go from, you know, it may blow up the first three times, suddenly it actually works, and then now you’ve got, how many string, you know, Falcon launches have there been in a row?

19:35 So awesome. Genius level insight. The issue with it is there’s a sliding scale. So early on and SpaceX is a great example, Falcon 1, Falcon 9 early days, very very focused on iteration and shipping unbelievably quickly. But when you look at a Falcon 9 launch when you’re putting astronauts and you’re taking astronauts to the ISS, there’s actually a very different safety profile and it’s a sliding scale and you’ve got to know when towards let’s just test it and it will break and that’s fine versus this can’t go wrong.

20:10 I don’t think China’s quite figured that out yet based on some of the recent rocket launches we’re seeing where the rockets are going up and they’re coming down back down crashing into villages. That’s a really clear example where China hasn’t quite worked out what the right balance between these two elements is and this is a sliding scale. So on next generation capabilities we probably want to err towards test, test aggressively, iterate, learn. But when you put people in the loop you probably want towards safety and reliability and it’s a very deliberate decision on that.

20:45 I also think on the regulation side for nuclear reactors the amount of money is egregious where it costs a few hundred million or a billion dollars of raw material costs to turn on a nuclear reactor and then there’s like 10x that on regulation and stuff. Do you think in order, let’s play out the scenario of the US has to win. How do we design something where we can actually have a shot at that? Do you think that we’re actually in a position today where we can do that or what needs to change in order for us to be in a position where we can do that?

21:22 This requires nuance. I think we’re playing the card right. I think we’re playing the game the right way. If we get too conservative and the US becomes Europe and regulation isn’t about safety, it’s a paper exercise which is by the way where I think many of the things today is that’s the way it is. It’s like I get to see, our customers at Flow are operating across EASA which is the European equivalent of the FAA and the FAA. And the FAA is still tricky but it’s an order of magnitude more innovation progressive than EASA. EASA is really there to have you filled in the paperwork, have you dotted the i’s and crossed the t’s even if the safety isn’t there. Whereas the FAA really do care about how are we making this safe, are we thinking about it the right way and what level is the dial turned to. There are different levels where humans are in the loop versus out of the loop and I think that’s the right call.

22:15 I would argue the space industry has a bit of an advantage compared to the aero industry where there is less regulation and that regulation is starting to come down a little bit and that’s enabling the industry to move faster. But where that careful balance is is tricky.

22:30 How would you also think about getting more founders into the areas? I think it’s something like 640-ish ship building capacity in China to one in the US. We have to have people actually take on the challenge of doing that kind of stuff. How do we do that?

22:50 The ship building thing is really interesting. And I think it shines a light on this meta point. And I call the meta point new game, old game. And people think that it’s about going back to the old game and just playing that game again, but playing it faster. And I would argue actually a lot of the game is the new game. So robotics are cool, but human robotics are cooler. Rockets are cool, reusable rockets are cooler. Old nuclear, SMRs are cooler.

23:18 We need to be able to be at par with the old game because honestly, we’ve gone backwards. We used to have a lot of these capabilities on shore. We’ve lost that capability. We got lazy. We got overly reliant and we got lazy and we need to bring that back. That’s necessary but not sufficient to winning the game. And winning the game is really about winning a new game. And this is where China is excelling.

23:45 Now, I think when you think about let’s play a sci-fi thing out here and let’s say money isn’t a factor and the US goes to war with China, is there a world in which that war is fought with human robots rather than people? Maybe. Right now people are cheaper unfortunately. There’s a deeply capitalistic thing here which I’m not okay with but is a reality. But if human robots fight human robots, who’s going to win? Who’s got the better robot? Probably not.

24:15 The way that that war is probably going to be won is when the drones talk to the human robots, talk to the autonomous submarines, talk to the autonomous tanks and that whole thing is working as one autonomous cluster and that cluster is being able to be better and the AI system on that cluster is better than the AI system on the cluster of China. And that’s the new game.

24:40 So this is why for a long time people misunderstood Anduril and they were like oh they’re just doing Boeing. And actually Anduril is a software company and the software company has a whole bunch of hardware components that need to be on the field that talk to one another but the underlying software is a really key innovation.

24:58 So going back to the question on if the war does happen how do we win? I think there are two key points. Key point number one is reliability. If China’s further ahead but all that stuff is breaking all the time, we’ve won. The second is the new game. And the new game for me is around how these things talk to one another and how software defines the hardware as opposed to a hardware first approach.

25:22 So it’s a little bit like the David versus Goliath situation, but you just basically need to make it so David can punch way above his weight. Yeah. The new game is all about how these things talk to one another. Not necessarily with humans in the loop.

25:40 Now, that’s a scary thought. If you go 20 years into the future and there’s human robots and the human robots can fight wars with drones and then you turn on autonomy and you say, have at it. And you need to make your own decisions cuz China’s turned on autonomy and they’re operating faster than we are. We’ve got people in the loop and that’s taking too long. So, China wins. If we turn on autonomy in drones and human robotics and something goes horribly wrong, then we’re in this weird sci-fi dystopian universe where we can’t really control these items. So it starts very sci-fi, but I think AI and the mesh layer above it will be as important, if not more important to the new game as how good your tank is.

26:22 And with those kind of thoughts in mind, how is that influencing how you’re acting day-to-day and what you’re focused on?

26:32 Well, there are two types of people. There are people that run towards the problem and people that run away from the problem. I’m someone that runs towards the problem. So I’m out here playing, fighting the good fight and doing what we need to do to win.

26:47 I think the really interesting thing about software kind of invading hardware is it again changes the game. So when I think about the history of manufacturing, I see it in three distinct eras. There’s what I call the analog era, the digital era, and the iterative autonomous era.

27:05 The analog era was era one of manufacturing. This is the Wright brothers, Apollo. This is the Manhattan project. It’s when we did things on pen and paper and we were kind of just working stuff out. Ballpark I think 40s to 80s and we were just getting off the ground. Our systems were low complexity but high enough complexity for us doing pen and paper and we were working stuff out. It was physical blackboards. It was drafting pen and paper drawing elements and it was physical test and everything was analog.

27:40 And then complexity increased and we entered the digital era. And the digital era we basically use software to design higher complexity systems in each constituent vertical. So we move from blackboards to spreadsheets and it enabled us to do way more complicated parametric models. We move from drafting tables to CAD and it enabled us to build way more complex geometry. We moved from physical test to simulation and physical test and enabled us to crank out the order of magnitude of testing and do way more tests way faster. And that’s kind of the era that we’ve lived in for the last 40 years.

28:18 When I think of the analog era, I think of Apollo. When I think of the digital era, I think of space shuttle. And again, that era sort of worked for a bit. And now that era is broken and we’re in era 3 right now. And era 3 is the iterative or the autonomy era. And complexity has again shot up by large magnitude. And this time what’s causing complexity to go up is actually software invading hardware.

28:43 So watches used to be watches. They’re now a computer on your wrist. And they do so much more than… they do so much more. And the complexity in the software is actually an order of magnitude more complex than the hardware or maybe even like 100 times. Totally. Cars aren’t cars anymore. They’re autonomous self-driving computers on wheels. Fridges aren’t fridges. Fridges have touchscreens on the front and they look at all of the things in your fridge.

29:15 So I think what we’re seeing happen is there’s this new era of hardware emerging where software is one of the key elements to what makes hardware successful. And when you’re in this era of 10x 100x more complexity, where software drives everything, then that’s what causes people to reevaluate how they work.

29:37 So in this third era and Apollo, space shuttle, I think of this era as really the Falcon 9 era. That’s kind of the image I have in my head of the defining product in this era. In this era we need to reinvent how we do engineering. It can’t be top down functional. It’s deeply cross-functional. It’s deeply iterative. Our systems are too complex for us to take a top down approach. We have to take a bottoms up approach.

30:03 And it’s really coined the term responsible engineer. So this is one of the key terms in El Segundo that you don’t really get anywhere else in the world. It’s this idea of an RE or a responsible engineer. And an RE is a mechanical or a domain engineer plus a systems engineer plus a project manager plus a manufacturing person all wrapped into one. And they take ultimate responsibility for the requirements and how we deliver against it. And the exact product that we build can change.

30:35 So going back to your question on how does that affect what I do day-to-day? Well, I think what we’re seeing is software invading hardware. When software invades hardware, complexity blows up and you need a new bottoms up automated way of testing these systems and an iterative way. And that’s kind of the backbone to why Flow exists and why I have a company.

31:00 And this kind of, I don’t know why I’m thinking about this, but Balaji has this basically response in the Lex Friedman podcast a while back that he did where he talked about, if you put a rat in a prime number maze, it just wouldn’t be able to figure it out. And most humans wouldn’t be able to figure it out. But if you look above, it’s very obvious on exactly how it works and there’s a simple way to go through this. Can you see the game in a different way because you’re able to see a top-down view of how all these companies work from the inside basically?

31:35 I think we have a unique perspective. I come from the old game. So when I grew up I studied mechanical engineering. I went to the old school companies doing the traditional way of doing engineering. I saw how fundamentally broken that was. I’ve seen the new way of doing engineering. A lot of the new way has been done on our software. And we’ve got to see up close and personally how these companies operate and build that into our software. And I think it’s as much a cultural change as it is a process change.

32:05 So let me use two words here and people conflate the two words and they misunderstand these two words and they think the two words are the same. One word is agile. The other word’s iterative. And I think both of these two things are deeply important to how the new wave operate.

32:22 So what is agile? What is it? And how they’re different? Well, a company like, let me use an example. A company like Radiant Nuclear or SpaceX are deeply iterative. And iterative means we take a bottoms up approach. We care more about what we can build and ship and produce in the short term and then we sort of quickly improve this on the fly. But they’re not necessarily agile and the reason for that is the requirements are fixed. They’re good requirements. We know that for SpaceX or for any launch company, it’s cost per kilogram and we want to minimize cost per kilogram and it’s very very clear what that requirement is. And that requirement isn’t going to change. At no point is humanity going to say we want to spend more money to put a kilogram into low earth orbit. And at no point are we going to say we want energy to be more expensive. We want this to be cheaper.

33:15 But there’s also a wave of hardware companies, and I think of like a Nuro or a drone company or Ring is a great example of this, where you have all of the hardware risk and complexity, you need this iterative approach, but if you don’t continuously question your requirements, then you don’t know if you’re doing the right thing.

33:35 Like for example, drones have become one of the most important elements in the war for Ukraine. People didn’t expect how important drones would be in this war and they’ve been one of the single biggest drivers for it. But when you start designing a drone, it’s unclear what this drone needs to do. What is the range of that drone? What capabilities does that drone need to have? Is it offensive or is it defensive? Is it got cameras on it or has it got payloads on it? Is it going to be autonomous? Is it going to be driven by someone? Is it big? Is it small? Is it $15 million? Is it 500 bucks?

34:10 And there is both the question of we don’t know what the requirements are yet and we need to go discover those requirements, but also the war changes. There are different eras of that war. The way that people play the game changes and the requirements on the drone companies change. So for a company like a drone company, you both need to be iterative, but you also need to be agile. You need to be continuously questioning your requirements, the top level requirements, not just the low-level requirements, and saying, are we producing the right core product?

34:42 And in your own journey, when you’re thinking about your kind of life and how you got here in the first place, what was that like? Cuz you didn’t start out wanting to build a software company. You started out building a rocket engine. And then you eventually figured out that building a rocket engine is hard and there’s this huge problem that is actually experienced by all hardware companies globally and you can go solve that problem. So in your mind, what have been the moments where you kind of questioned your own requirements on what should I actually be doing in the first place? Does it make sense to just completely pivot to this other thing because it’s clearly a problem?

35:22 I think a macro view on this is founders are pretty good at iterating with data. They’re pretty bad at changing their philosophy on are we asking the right questions at a fundamental level. And I think I’ve been lucky enough to be in a company where we’ve worked for eight years really really hard and every two years we’ve had to question the core philosophy of the company and why the company.

35:47 Do you want to go through those? Yeah. Let’s talk about the last eight years. So I’ll tell you the story of the company. I’ll do the slightly longer version because I think you’ll see the assumptions we had, those assumptions being broken, and actually we’re playing a different game and the game that we’re playing is completely a different game.

36:08 So we started the company, I was 21 years old and we just said, hey look, the world is wrong about how to develop hardware. We need a new way of developing hardware. We’re just going to go do it. The fastest way for us is just to go do this ourselves and we need to earn money. We don’t need to earn a lot of money. We need to earn some money to be able to pay for the bills. So we’re going to just do a consultancy and we called ourselves the rocket company.

36:35 And as I said earlier we were the fastest design consultancy in the world for hybrid rocket engines. The fastest people could do it in 12 weeks, we could do in 2 and a half hours. And in that phase one we said look what we’re going to do is we’re going to be the role models and we’re going to invent this new way of working and it’s going to be so much more productive that everybody else in the world will kind of need to copy us to be competitive. We’re going to do it for rockets, rocket engines, and then somebody’s going to do it for wind turbines, and someone’s going to do it for nuclear, and then over the course of 20 years, the entire world will change how they work.

37:08 Our mission’s always been the same. This is really interesting. Our mission has always been to reinvent how to develop complex hardware, but how we’ve done that has changed.

37:20 So we did this for a year and at the end of the year, we realized people want to develop in this way and the reason they’re not able to develop in this way is the tools don’t exist yet for them to do that. And if the hardware engineering industry is going to change how they work, we need a new type of platform and a new type of software to enable this way of working.

37:42 So we made the mother of all pivots. We pivoted from a hardware company to a software company. And we said we’re going to build this core technology. And NASA tried to build this for three decades. They called it the intelligent census environment for one decade. They had different names for it. I’ve read every single paper NASA’s ever written on this over the course of 30 years. And they took three different stabs at it.

38:07 And we said, look, if we can build a software and we can build it in a way that’s beautiful and elegant and people can use it on day one without training, then it’s going to be so much more productive that people will need to change how they work. So we went into this little man cave and I was 22 at this point and we raised a bunch of VC money to go fund it. And we said to ourselves, we’re going to build a DeepMind for hardware. We’re going to build this new type of research lab. We’re going to invent not just the tools, we’re going to invent the paradigm for how to work in this way. And then we’re going to take it to companies like McLaren and Rolls-Royce. And we’re going to show them how much faster it is. And they will have to adopt it. Like if we’re able to invent the technology, humanity will change. And the company might die, but we can just open source technology and humanity will be… it will still be very valuable.

38:58 And we went into the man cave and we worked insanely hard and we made the breakthroughs. And we not only worked out this new paradigm of doing engineering, we also built the tools to enable that. And we came out of the man cave and we took it to McLaren and Rolls-Royce. And I remember super clearly we took it to the directors of these companies and they said we were here with the last platform change. We were here when we went from drafting into CAD and now we get to be here when we go from CAD to this and it’s a new way of working and every vehicle we ever build in the future will be built this way. There’s no way it doesn’t happen.

39:37 And at this time you were still based in London, right? And so that was kind of knocking on their doors. It was right down the road sort of thing. Totally.

39:47 And we got into these companies. We were called the engineering club at this point and we were so excited. We’re like we get to change the way these companies work. And then we got into these companies and we saw the same core pattern which is we get into the companies, they were overwhelmed with this whole new approach to doing hardware. They were 3 months behind on the current vehicle and they didn’t know where to start and we spun out every time.

40:12 And it was this huge realization that we had. Realization one we had from the rocket company to the engineering company is you need to have good tools to enable people to work in this way. But then realization two we had is that’s not enough. The real problem isn’t just software, it’s behavioral change. If you have the right software and no one changes how they work, then you haven’t actually changed the industry.

40:38 And we had these deeply technological motivations before. We just build the tech, we open source it, kaboom, the world is different. And what we realized is that wasn’t enough. There needs to be this education and this whole ecosystem around it to enable… somewhat of a movement. Exactly. We need to build a movement.

41:00 And the last point we realized in this is people don’t understand new things. This is really like one of the most important learnings I’ve ever had. The first cars weren’t called cars. Do you know what they were called? I actually don’t. I think the first cars were even electric. They were electric. The first cars weren’t called cars because people didn’t understand automobiles. The first cars were called horseless carriages because people didn’t understand the idea of a car, but they got carriages and they got it has to have a horse. And this is a horseless carriage, a carriage plus.

41:38 I think that’s a lot like the autonomous cars of today. Like they aren’t really cars, they’re just autonomous vehicles. That’s exactly right. What do you call an iPhone? People call it a phone. It’s not a phone. It’s a touchscreen computer with a phone app on it. 1% of the value I get from my iPhone is making phone calls. 99% of the value is all this new stuff that I can do on the new platform that I couldn’t do before in my sort of like Nokia phone.

42:08 But people are bad at new things. So what we had to do is we had to take this new platform, this new way of working and we had to hide it and disguise it into a workflow that people understood and people knew and that workflow was requirements management.

42:25 So I never thought 6 years ago I’d be building a requirements management company. But that’s what was asked of us. So we had all these European companies. These European companies were deeply wedded into working in this way but they didn’t have the use case and they didn’t have the culture to actually transition over. And then we had this one tiny customer in the US and they were five people at the time and they were called Stoke. And Stoke are a rocket company. They make reusable rockets and now they’ve raised a couple hundred million. It’s a couple of years later but back then there were five people. I think they got to the first reusable rocket engine on 10 people. Yeah, amazing team. We love working with Stoke.

43:10 And they said the same thing that McLaren said. They said this is the future of all of engineering. In 10 years, all engineering will be done in this way. In the same way, if you turn off GitHub today in software, the entire industry will crumble. In 10 years, if you turn off Flow, the entire industry will crumble. But you’ve got to let us put our requirements in there.

43:35 And we were like, what are you talking about? We’re building the future. We’re building this whole new platform, this whole new way of working. It’s integrated modeling. It’s autonomous git-like branching, and you get to work in this new way. And they were like, yeah, yeah, we get all of that and we want all of that. But if you don’t solve a problem for us today, then it doesn’t make a difference. And the first problem we have and the most acute problem we have in the organization is requirements management. We have requirements. Requirements are the source of truth for how we operate. Every single design and variable and model and test need to tie back to requirements and the requirements change on a continuous basis. And if you build this platform, but you build the app on the platform and the first app you build is requirements management, then we’ll use it for requirements management. We’ll come for that thing and then we’ll realize a lot of the new stuff is why we stay.

44:30 And we tested it out and we said we don’t want to do this. We’re a new company. We believe in products and technology. We don’t want to do this old use case. We tried it out. When we tried it out, we had a number of design partners. They all started to use it in production the same week for requirements with live data. So the thing that you didn’t want to build was the thing that was working. The thing we didn’t want to build was the thing that made us successful.

44:55 And by the way, the really ironic thing is they all started to use it for requirements. And when they could understand requirements in this new way, they could understand the platform. They could understand the whole new way of working. So the way that our customers think about us is requirements management plus. The way that we think of ourselves is we’re building this new type of development platform that didn’t exist before and requirements is a Trojan horse that enables us to get success.

45:22 And that’s what made it successful. And I think the third fundamental philosophical change we had is it’s not good enough just to build the technology. It’s not good enough to build the technology and the paradigm. You actually need to solve a problem that people have today as your starting point. And then when people trust you to solve that problem, they’re going to come to you for the next problem and the next problem and the next problem. And over time, you’re going to Trojan horse yourself into their tool stack and they’re going to start to use your tools to work in this new way.

45:55 It’s not an easy thing to make the call to go into a new direction. And even when you’re talking about some part of the software is something you don’t even really want to work on. Then suddenly people, that’s the most valuable thing. That’s the beach head.

46:12 When you were making those decisions, what was running through your mind? How did you decide to kind of go in different directions?

46:22 So the story I tell myself is the more ambitious the mission, the more difficult it’s going to be, the more acceptable it is to take time to build each layer of the stack. Cuz if you think about Flow, we’ve been on the same mission for eight years. I have set my entire life purpose on this one mission, which is reinvent how humanity designs complex hardware.

46:45 And we started out the paradigm layer. That was the rocket company. And then we did paradigm and technology. Then we did paradigm technology and product. Then we did paradigm technology and product and beach head app. And basically and then by the way we did all of that stuff and it still didn’t work. And it still didn’t work cuz we were selling to ancient European companies that didn’t really have an impetus to do things in a new way cuz they’re still building the old way.

47:12 So we had all of that and then we had the macro move with the El Segundo market to actually is there a reason why we need to do it in this new way and is there a new culture and a new set of people that are going to do that.

47:28 So we had to invent multiple layers of the stack. Each layer of the stack, the paradigm layer requires a different skill set to the technology layer, which requires a different skill set to the product layer, which is like UI/UX, which requires a different skill set to the evangelism layer, which requires a different skill set to the workflow layer.

47:48 So we’ve had to get good at every single one of those layers. But I think that is what makes Flow a really special company. We’re not just inventing tools. We’re inventing tools, paradigms, processes, evangelism. We’re inventing this whole new way of working and that is what makes it go from here to here. That’s what enables us to do the crazy things that we’re doing.

48:12 And I am sure 3 years from now there’s going to be three more layers. There’s going to be an AI layer to this. There’s going to be some sort of marketing evangelism education layer to this. It’s not just good enough for a new wave of El Segundo companies to do it. We actually have to help the old school world convert how they’re doing. We actually have to train a whole new wave of engineers that won’t necessarily work in this way to help do this. For us to be able to make Flow successful, to reinvent how humanity develops hardware, it’s going to require education. And that I love. That’s why I get up out of bed in the morning. I love the game.

48:52 And I think that kind of goes back to it’s not reinventing the mission or what you were going after, but it was understanding that the way to get there or the path to get there may have been slightly different than you were expecting.

49:08 I think for a long time, and this comes from high testosterone 21-year olds, we were, the internal monologue for us was they’re wrong, we’re right, and they don’t get it and we’re going to show them and we’re going to show them at such a scale that they have to go, we’re wrong. And it was very antagonistic and it was let’s show the world what the right way of doing this is and we’re going to lead from the front.

49:35 And I think as I’ve grown and matured and as the company has grown and matured, we now have a different approach, which is rocket scientists are actually really smart people. These people are amazing and they’re on the front lines. And in this story, they’re Mario and we’re the little powerup and we turn Mario into Fireball Mario and we help Mario be even better. And we’re not the heroes, they’re the heroes. And we’re in this game to serve them. And if they ask something of us, we’re going to rise to that call.

50:08 They’re trying to save the princess. And if you can give them a power up to shoot fireballs, that’s great. But they have to figure out how to actually use the power up. Yeah.

50:18 And this is why when we were selling into Europe back in 2018, 2019, there was no conceivable way for us to make that game work because there wasn’t the appetite for them to change how they worked and to completely transform and change the culture and the process and the tools. But what you’re seeing happen right now, and this is why this generational change is so important, is it’s a whole new way of working. It’s a whole new culture, and this is the culture that will be successful.

50:48 You mentioned kind of being written off by people and how Mark Zuckerberg has this talk I think with Paul Graham from I believe like 2012 and he talked about wanting to be like, there’s different states and sometimes you’re underestimated, sometimes you’re overestimated, but his favorite state of being is where people don’t believe that he’s going to win or that he can make the thing work and it’s actually better for his mental state to be in that area. How has that kind of been for you?

51:22 So there’s an assumption in your question and the assumption is the scope. So if I say Zuck is overestimated, well, it depends on what you’re trying to do. If Elon was trying to build a rocket engine, I think people can go he can probably pull it off. As a software founder, if he says I’m trying to put people on Mars, then suddenly the scope is so much bigger that people are like obviously this is doomed to fail.

51:48 For me, I’m just going to increase the scope until I am underestimated. And that’s the fun of it. The fun of this is the way that I think about this and how I thought about it from a pretty young age is there are two parallel universes. One universe in which I wasn’t born, one universe in which I was born. And the net positive difference between these universes is my impact as a human being. And my life purpose is to maximize the positive impact that I can have in this.

52:20 And for me, the thing that I can commit my life to is redefining how the design of every physical thing in the world is done. That’s a pretty cool thing to leave behind. And if the entire industry goes from this to this, that’s pretty huge. But if the entire industry goes from this to this, that’s really really meaningful.

52:42 If we look forward 200 years and we stop thinking about satellites and SMRs and human robotics, but we start thinking about the Starship Enterprise and it’s not a thousand parts in the bill of materials, it’s a billion or a trillion parts in the bill of materials, to pull that off will require more engineering hours than there are people in the world right now. It’s impossible to do. So it’s going to require us to again reinvent how we do engineering.

53:15 And by the way, I’ve got a pretty good… So I said there’s era one of engineering which is manual analog. Era 2 is digital. Era 3 is iterative. I’ve got a pretty clear image of what era 4 is. What is era 4? Era 4 is computational and generative. We’re not going to have somebody go into CAD and draw the 15,000th valve for a slightly different condition in the Starship Enterprise. A lot of that’s going to be generated. And that’s going to be this new computational approach to doing hardware. But we’re not there yet, and we don’t need to be. We need to win era 3 and we’re right at the beginning of the S-curve in era 3. And this era is going to last 20 or 30 years. And then we’ll do era 4.

54:02 And I’ve got a pretty good bet on era 5 as well, which is more nano technology than it is about like if this era is around software starting to enter hardware, that era is going to be software is hardware. And it’s all kind of the same thing. But we’ve just got to take where humanity is today and take the next step and help them to do that. Take the next step and help to do that. And we’ll just continue to take steps.

54:30 And this is also something where you just can’t do a massive leap from era 1 to era 3 or era 1 to era 4 and you kind of have to go through this process. Yeah. You can’t skip steps. You can just do the steps faster.

54:47 How are you trying to kind of shorten the time frame of people adopting this? If you are really trying to get humanity to go from this to that to that, I imagine that there’s something in the back of your mind where you’re saying if we can do this one day sooner and then that one day compounds and compounds and compounds, it has massive impact years in the future. How are you thinking about that?

55:15 So let’s go back to the Mario analogy. I don’t think we’re Mario in this analogy. I think our customers are Mario. So the people designing human robots and reusable rockets and self-driving cars, they’re the heroes. So the question is there is this macro shift. There’s this tiny new wave of cool new startups that are designing next generation systems in this crazy way and then there’s the old school legacy dinosaur. And the dinosaur is trying to kill the new wave. The dinosaur’s only means of survival is to kill the new wave so that they aren’t obsoleted.

55:50 And success for the new wave will become so productive and so powerful that they make the dinosaurs look like dinosaurs. And I think when you look at EVs, we’ve had this. When you look at reusable rockets, we’ve had this. Even seven years ago, Europe and the founder and the CEO of Ariane Space, which is a European launch, was like reusability is a pipe dream. It’s never going to happen. And now SpaceX does 95% of mass to orbit. Like 95%.

56:22 Is there even a historical analogy of something like that happening where there’s a single organization, not nonetheless company, but organization that is fundamentally moving an industry forward like that?

56:35 I think it’s probably the norm. I think if you look at Rockefeller and you look at the rail and you look at the East India Trading Company, if you look at any of these industries, typically one team does it the best.

56:48 But to go back to the point, I think what we can do to make this work is either the El Segundo next generation hardware companies will have their shot and they’ll screw it up. If they screw it up, this game’s over and we retreat 30, 40 years. Boeing says, look, we told you so. These new kids tried and the new kids failed. Or the new wave will become the SpaceX for nuclear and the SpaceX for aero and the SpaceX for robotics. And you’re going to see this whole new wave of companies emerge. And these companies will become the mass market.

57:25 And in the same way when you look at software, this exact story played out. You had IBM and you had Oracle and you had Microsoft and there were these huge Goliaths that operated in a very certain way. And then you had these new upstarts like Google and Facebook and Twitter. And these new upstarts were doing engineering, developing software in a completely new way to the old way. And those companies became so successful that they became the Fortune 10. And then eventually the old boys woke up and they said maybe we should adopt GitHub. Maybe this idea of a Gantt chart isn’t the best idea and maybe we need to become wise and they eventually came into the new bubble.

58:08 So what we can do to be successful is we can just get this little bubble of an emerging industry. We can go all in on there. We can make those guys as successful as they can possibly be. We can give them the best possible shot of success. And if they become successful then the industry moves.

58:28 And I also liked the way that you described getting the product into customers hands in the first place. You’re not trying to sell an entire organization. You’re trying to sell an individual. It’s a little bit like Slack almost where you basically just have people using the thing and then suddenly because they’re using the thing, the organization adopts the thing.

58:53 And I almost imagine that it’s a little bit of a situation even on kind of like if you think about taxes and government, if you go up and up into a bigger and bigger organization, it’s less and less effective at allocating dollars. Whereas a smaller individual person is really good at allocating $10 whereas a massive trillion government can’t allocate the same $10 in the most effective way.

59:20 It’s a little bit similar in that sense as well that instead of trying to get an entire organization to adopt, you basically just try and find that one person that’s allocating their own time and then figure out how can I make that person more efficient and then eventually the entire organization figures out this is the right way to do things.

59:42 I think there’s a meta point here. When you look at the people that we serve, we serve roboticists, we serve rocket scientists, we serve people designing nuclear reactors. We serve the most creative and innovative young engineers of life. If you look at the tools that these engineers use on a day-to-day basis, it’s horrific. The tools in our industry are stuck in the dark ages.

1:00:08 If you look at Siemens, you look at Dassault, you look at a lot of these major players, the software was built and conceived in the 70s and the 80s. And when you look at the software today, you can absolutely see that it hasn’t changed. In systems engineering, the market that we serve, the monopoly, the single biggest player is IBM. IBM is the king in our industry. And it’s crazy to me that we have this dichotomy between the most impressive frontier technologies and technologists in the world are being built on this archaic ancient software that needs to be in the bin.

1:00:42 And you look at any other industry, let’s go back to software, gaming, chip design, even graphic design. Any of these industries, they have amazing software that are built to enable these engineers and designers to do more. And you look at our industry and it’s completely opposite. And the tools are actually getting in the way of these engineers being able to do stuff.

1:01:08 So then you ask the question, why? Why isn’t the free market working? Why are these amazing companies and engineers stuck with this really crummy software? And the reason why is our industry, the software for hardware market is a very deeply entrenched oligopoly. There are four big players in the market: Siemens, Dassault, Autodesk, PTC. And those are the four big CAD vendors. And the CAD is the sort of like the programming language equivalent. It’s the thing that every one of these companies are centered on. And then once you have that monopoly, anything that integrates well with that is sticky.

1:01:48 So these companies can afford to have really shitty software that doesn’t work half the time that people don’t understand how to use and still make seven, eight figures a year from leeching off of our customers because there is this generational change happening from waterfall to agile. It rewrites what companies need out of their software. It rewrites the core workflow and for the first time in maybe 40 years there is an opportunity to rebuild the entire stack and to say what would the best CAD PLM whatever system look like if it was built in the 21st century and it actually helped engineers do more not less.

1:02:30 And I think that is the opportunity that is emerging today and it’s not just Flow in that market. There is this whole new wave of software for hardware companies emerging. What I see being the most important thing in this however is taste. And even a lot of the new wave companies I think there are some that are exceptions that are really fantastic at this but I see really crummy software built.

1:02:58 I’m a mechanical engineer and when I go to any software that is even a little bit crummy, I turn off, I go to Excel and I just do all my work in Excel and I’ll translate it to the software later on. There are very few companies that can do what is the next generation workflow and what is an incredible experience that works for the end user, not for the decision maker, not for the buyer, for the end user that’s actually doing this and make that work collectively.

1:03:28 I think it’s a decision as much as it is a differentiator. And I think it’s one of the things honestly we do the best. We not only work out what the new workflow is and we educate people and we help engineers operate in this new way, but we believe at a very deep level that if we’re going to build software and this software is going to change how people work, they need to adopt it and they need to love it and it needs to be an incredible experience, not just a good enough experience to get by. And that’s why when you look at Flow versus traditional software for hardware companies, it’s an entirely different approach to UX, to quality, to product design, to intuitiveness.

1:04:08 You were talking earlier about the types of people that you kind of look for and the two different types of people. Can you kind of explain how you’re finding those folks? What is your philosophy for finding talent?

1:04:22 So Flow has a very very different culture to almost any company that I’ve seen. And the way that I describe it internally is there are two types of organization. There is the army and there is the special forces. And the tools and the processes and the cadence and the rigor that works for the army rank and file people, they have their spot, they fit into the slot and they do that thing very well. Versus the SAS or the special forces is entirely different.

1:04:52 The right process and cadence for the army absolutely turns off the special forces elite and the right tools and the process for the special forces elite doesn’t work for the rank and file army person. And we made a very deliberate decision on our culture which is we want a special forces culture not an army culture.

1:05:13 And our special forces culture has enabled us to do crazy amounts. So I think the industry incumbent we compete with is a company called Jama. Jama’s team is 3000 and we win head-to-head on every deal we take together. Head to head, we win every time. When we raised our series A, the entire organization was eight people and we were aggressively profitable. We couldn’t hire quickly enough and we have the best product in the industry by an order of magnitude.

1:05:45 And the reason that we’ve been successful at that is we’ve built this special forces culture. So what is it that we look for in the special forces culture that is different to traditional hires? And the framework that I built is something I call zeroth order, first order, and second order people.

1:06:07 So zeroth order people I’d say are 80% of people. It’s school, college, job, job, promotion, job, retire. And it’s the kind of default career track. It’s people that are looking for a 9 to 5 and it’s people that want balance. And there’s absolutely nothing wrong with that. That is the vast majority of people in the world and I think that the world is better off for it.

1:06:32 And then occasionally I’d say maybe 15% of people fall into this first order category. And the first orders are your classic high risers. They’re your McKinsey, your investment bankers, they’re your Googles, your MBAs. It’s a slick back-haired person that did really well at school and then did really well at college and then went to the prestigious place. And it’s a little bit like an ambitious version of the zeroth order.

1:06:58 I think these people are great. I think these people are high performance people and when you look at CEOs of large companies, they’re typically first order people. I’d say the world runs on zero and first order people. That is what is required to keep the world running.

1:07:15 But I think the people that change the world are second order people. And second order people and first order people are very very different. I’d say only like 5% of people are second order people. And second order people, it’s both a talent thing and a decision thing that you need both of these two. And what second order people are are your exponentials.

1:07:38 So what I look for in second order people is the amount of personal development at every stage in every role increased. So you learn this much in the first role and then this much in your second role and this much in the third role. And I said it’s both talent and a decision. The talent thing enables the rapid personal growth. The decision thing is typically when you are in a job you cap out at a certain level and you can’t take the next step up because you’re in an organization. And to take the next step up, you actually need to take a local hit. You need to take a step down in pay, in location, in quality of life, in seniority, in job, whatever it is.

1:08:20 You have to be willing to take a short-term loss, a slightly short-term loss in order for a longer term, much higher slope. And that’s the key thing. They have to take a local hit for a higher slope. And very few people make that trade. Very few people are able to go, I’m going to take a worse salary and have a longer commute and do more hours, but I’ll be at a higher slope 2 years from now. Very, very few people can do that. But the people that change the world are those people.

1:08:53 And when I think about our special forces culture, our special forces culture is built on finding these second order people, getting them into Flow, and then having an incredibly high bar.

1:09:08 I wonder if that’s even trainable. Is that an internal thing when someone has that second order mentality? Or is it something that they can learn from past examples of the optimal long-term move is taking these short-term hits in order to increase that slope?

1:09:28 It’s probably both. The way that I think about it is who are the people that compete in the Olympics? Who are the people that are the best in their game and how do you become that? And part of it is I have a good coach and I see the best and I imitate the best and I learn and I work insanely hard. But part of it is also innate and it’s people like I think I remember Usain Bolt. Usain Bolt did the world record for the 100 meters and just before he did the world record he had a box of 12 chicken McNuggets.

1:09:58 So the best of the best of the best aren’t just best on their best day at the top 1% of sleeping really well. They can have 12 chicken McNuggets, go into the Olympics, and set the world record on 100 meters. That still hasn’t been beaten, I believe. So it’s I think kind of both.

1:10:20 I don’t know if it was Michael Jordan or there’s some very famous, it was either Michael Jordan or Kobe. And I think some interviewer asked him, what were you doing when you were a kid? And his response was basically, well, I was playing baseball. So until like age 12, the guy was playing baseball. And then he decided to basically go into basketball instead. And he took the same sort of practice regimen that he took at playing baseball towards this new thing and he was able to hit the growth trajectory enough to actually make it to the top.

1:10:58 Yeah. I think it’s much more about attitude than it is about talent. I generally think if I could trade myself 20 IQ points for 20 hunger points, I’d make that trade.

1:11:12 I love that. I think Paul Graham even has that line where he basically said if you take two people or if you imagine people 100 out of 100 on intelligence, 100 out of 100 on determination and you start taking away determination, you basically get this genius taxi driver that doesn’t really do anything. But if you take a hundred out of 100 on intelligence, 100 out of 100 on determination, and you start taking away the intelligence side, you basically eventually get to a point where someone is not like making the next video game or whatever, but they are basically running some massive trash collection company. They’re massively determined. They eventually succeed, but it’s on something that doesn’t really matter.

1:11:55 Of the two, the person that ran the massive trash company versus the taxi driver, who did better for humanity? And I’d argue the trash guy, but by far because he built something that the world needed and he contributed.

1:12:13 And I actually think IQ is really overrated. You want people that have raw smarts, but I feel the smarter you are, the more you see the possibilities of things that can go wrong and the more risky you understand a situation to be. But people typically overindex on what could go wrong as opposed to what could go right and they don’t see the other half of it. So if I could narrow the scope, reduce IQ points in my team and increase hunger, increase drive, increase wanting to win, I think that’s the trade you make every day.

1:12:50 How do you think, from a perspective of in order to kind of eventually get the outcomes that you, like the massive outcomes, you have to really ask the right questions of yourself and kind of figure out what is worth working on, what is worth going after and spending a huge amount of your life making something that doesn’t exist in the world exist? How do you internally and then just generally think about asking yourself the right questions to eventually get the outcomes that you want?

1:13:22 I would challenge the question. I think it’s the wrong question. Let’s say I spend a year, I take a gap year and I spend a year thinking what is the highest value thing I can do for the world and then I realize that actually if I think about this a bit longer the outcome could be wildly different. So I spend another year thinking about what I could do with the world and spend another year and I eventually work out that the right thing I can do for the world and the world changes and suddenly there’s a whole… so I need three more years.

1:13:53 And if you look at Europe, I think Europe is a really great example of this where they’re so intellectual, they’re so bright but they kind of get stuck in this… So I think the right actual approach is what is the most I can contribute to the world today? Is it being a nurse or is it I’m naturally good at this thing over here and I could do twice as much of that. And then if you locally optimize but take data from the world, I think it’s very easy to… I think Paul Graham talks about this. It’s very easy to come out of a local optimum and to have contribution and then ask yourself continuously can I contribute more and then just to reassess and iterative cycle than it is to answer the question like what is the best thing I could be doing with my life and not do anything at all.

1:14:42 Yeah, it’s possible that I kind of misphrased that cuz the way I think about it is very similar to Brian Armstrong who said a while back action produces information. So you need to be basically taking action. And I think it’s very important from the question perspective. It’s just so natural to me to obviously you need to be producing work and doing things while you’re also constantly running this loop in your head of am I doing the right thing right now? Is the thing that I’m actually doing getting me directionally correct closer to the thing that I want to be achieved?

1:15:18 If you go back to the hardware stuff, this is exactly what agile iterative is. It’s do stuff and then just continuously check and refine and make sure you’re on the right trajectory and if you’re not, we’ll course correct, but we’ll do that five iterations in with a whole bunch of data and a whole bunch of flight heritage that we didn’t have before.

1:15:40 Final question. What’s the hardest thing you’ve overcome?

1:15:44 I don’t understand the question. There’s this parable that changed my life and it’s the parable of the Chinese farmer. Have you heard it?

1:15:55 So back in the ancient feudal era, there’s this village in China and the village is very poor and the village is 100 people. And in the 100 people there is a farmer, an old wise man, and a young boy. And one day the young boy goes out and he gets lost. And the whole village goes, oh no the boy is lost. And the farmer says, we’ll see.

1:16:18 The next day, the boy comes back with a donkey. The whole village goes, how amazing the boy’s found a donkey. And the farmer says, we’ll see. The next day, the boy gets on the donkey, goes for a ride, falls off, and breaks his leg. And the whole village says, oh no, how terrible. And the wise man says, we’ll see.

1:16:40 The day after that, the village goes to war. And all the young boys and the men go out and fight and die. And the one boy with his broken leg stays behind and lives. And the whole village says, oh, great. And the old wise man says, we’ll see.

1:16:58 So, I think people get very caught into the day-to-day difficulties of life. I’d say every single challenge we’ve had from rockets to software to paradigm to location to Trojan horses to evangelism has been part of the journey.