Jarek Rzepecki from Monumo: Motor Simulation and Optimization...For Now

Rare earth magnets, AI data center energy demand, and electrification are colliding in one place most people ignore: the electric motor. I sit down with Jarek Rzepecki from Monumo to get practical about what it takes to design motors and powertrains when costs, materials, and constraints can shift fast, and when “just optimize the motor” is never the whole story.

We dig into why system-level optimization matters, how a change to one component can cascade through the entire design, and why engineers can’t realistically brute-force the search space as parameters multiply. Derek explains how physics-informed AI, machine learning, and simulation can work together to explore designs faster, including approaches that reduce reliance on rare earth magnets while keeping performance targets intact. We also break down the motor landscape in plain terms: permanent magnet motors, wound rotor designs, and magnet-free reluctance motors, plus the real-world problem of torque ripple and what it does to noise, vibration, and durability.

Along the way, we connect the dots to robotics actuators, drones, generators, and the broader sustainability angle, because improving efficiency on the generation side and the consumption side can move the needle at global scale. If you like engineering software, multi-physics simulation, FEM, and the future of AI for engineering design, you’ll get a lot out of this interview. 

Transcript

Welcome And A San Francisco Hello

Roopinder 0:07

Hello and welcome to FoDES, the Future of Design and Engineering Software Podcast. My name is Roopinder Tara. On the show, we will have guests that will discuss tools and technology that engineers will find interesting and useful. Jarek, nice to meet you. Yeah, nice to meet you as well. Yeah. Hey, thanks for being on the show. My pleasure. Jarek, where are you calling from? Cambridge, correct?

Speaker 3 0:31

No, I mean I'm normally based in Cambridge, UK, that's correct. But today I'm in San Francisco.

Roopinder 0:36

Are you? Yes. I will explain your LinkedIn profile, which had me confused because it's a Golden Gate Bridge.

Speaker 3 0:43

Oh, yes. Um, that was actually I took that picture last time I've been here, so six months ago or so. Went for a walk, as you do, right? Um, and obviously there is this very beautiful beach, and you can see the Golden Gate Bridge. So, yeah, took that picture.

Roopinder 0:58

Um yeah, I'm on the other side of the bridge. Did you know? Oh, are you?

Speaker 3 1:03

Okay, I actually I did cross the bridge a few times when I go for a run. Because usually when I'm when I'm in San Francisco, I go for a little run, and then I usually run across the bridge. But Lichi, go across the bridge, turn around, and then go back. Come back.

Roopinder 1:16

Okay, yeah. So that's over. You go over the bridge, that's Marin County. I I've lived in a lot of places around the world. Not a lot, a few. I think that's my favorite one. It's beautiful. You should go out and blossom. Yeah, yeah, yeah.

Speaker 3 1:32

One thing that I really want to see around here is the you know, the the the red pine, the red wood forest. You know, that's I haven't, and this is totally on my to-do list. I need to go and see that.

Roopinder 1:42

How long are you staying?

Speaker 3 1:44

Until Saturday.

Roopinder 1:46

Saturday. Okay, all right. All right, well, very good, very good. Uh well, and you're a runner too?

Speaker 3 1:52

Uh yes, although not not a crazy runner. I do run casually. Uh, I do like the long distance running, so uh more like you know, tens of kilometers, especially off track. I really like running sort of in the wild or not on the not on pavement.

Roopinder 2:09

Okay, all right, yeah. San Francisco has its marathon, as do a lot of cities. Uh, but it also has probably one of the oldest races in the world. You have to come back for the beta breakers. It's called Beta Breakers, and it's a short race, it's not like long distance. So if you're running, if you're training for long distance, you could probably run this one easily. It's only seven miles, it goes from one side of the city to the other. But it's so much fun, Derek, because it's people dress up for it, they wear costumes, right?

Speaker 3 2:40

It's it's common like in London Marathon, people do the same thing, like they dress up in a crazy costume and then they land uh land around London Marathon. Yeah.

Roopinder 2:50

All

From CAD Copilots To Today

Roopinder 2:51

right. Hey, uh, our interests may have crossed uh many years ago. I looked at your background, and you actually did some programming for the synchronous technology in Solid Edge, right? Solid Edge is a I'm a big one.

Speaker 3 3:06

I did, yes. I worked on um uh a toolkit called VTK for them, uh, which was quite interesting. Uh it was a toolkit that, you know, if you work on any cut-cut file, right, and you want to modify uh you move things around, right? But then very often you want certain properties of your design to be preserved as you modify it. Uh, you know, to give you an example, if two two cylinders are coaxial in your design, when you move one of them, likely you probably want to move the other one as well and maintain the coaxial property between the two, right? Um, so that was basically developing software that would detect uh properties like that, detect properties that are likely uh to remain uh fixed when you modify the cut uh design and then automatically update them as you modify it, right? Obviously, the user always can overwrite it, right? That's clear. Um, but it was quite interesting because it was very early stages of developing algorithms and yeah, like tools that now one would classify as copilots, you know, like very people talk about copilots a lot nowadays in many different branches of engineering, software development, etc. And that was like quite an early attempt at doing that, right? It was software and algorithms that would help the engineer do their work quicker in an intelligent way. So, yeah, that was quite quite interesting. I enjoyed it. It was uh uh in Cambridge.

Poland To Germany Through Physics

Roopinder 4:33

I'm always interested in people's background, especially when paths intersect. A little bit more about your background. You are initially, so I'm foreigner, I'm foreign to the US, so I always ask about people's origins. You're originally from Poland, right?

Speaker 3 4:48

Correct, yes, yes.

Roopinder 4:50

Okay, tell me how to pronounce your last name.

Speaker 3 4:53

That's the tricky one. Uh so rz in Polish is pronounced z. Like you can think like the French z, you know, Jean-Claude. Uh, that's the same sound. Um, okay. It's tricky. Like, you know, if you don't know it, you would never pronounce Rz as Z, but there you go.

Roopinder 5:10

Thank you, thank you for that. Uh and then you went to you studied in in Germany?

Speaker 3 5:16

Yes, so I did my master's in theoretical physics in Poland. So that was uh my master's degree in in a town called called Torwin, which by the way is the hometown of Nicolas Copernicus. So I did my uh my master's in hometown of Copernicus, and then I did my PhD in Germany. So I found my PhD I moved to Munich and I did my PhD in in suburbs of Munich called Garshing Forschung Centrum, where I worked on astrophysics. Uh so my uh yeah master's was in physics, PhD in astrophysics.

Roopinder 5:48

Uh okay, all right. Now is that when you worked with uh Siemens to no, that was

Why Leave Academia For Industry

Roopinder 5:55

after.

Speaker 3 5:55

So uh basically after I did my PhD, um, I've decided that I don't want to stay in academia and I would like to work in industry. The main reason being that I really enjoyed seeing direct impact of what I do, and in astrophysics, that's hard. It's very interesting research. Um, I mean, I have very high uh sort of I think very highly of scientists in general, you know, people who just spend their times pushing the frontier science, it's great. But like for me, I really lacked a little bit of seeing the actual direct results of what I do. So I decided to left uh to leave academia and join industry. And the first job I actually had was in video games, actually. You know, software development was always a hobby of mine. Uh, since I was like 10 years old, I was playing in Assembler, you know, writing like special like graphical effects on you know, like 16-bit bit computers and all of that. So it was always a passion for me to do that. So then I thought, you know, if not academia, then what kind of industry? And I thought, you know, developing video games is as cool as it gets, right?

Roopinder 7:03

I understand it because a lot of the graphical applications are they are cutting edge on video games, absolutely, and uh, but it's seem for I have to explain to people that might be listening, it's a big jump there between astrophysics and gaming, right?

Speaker 3 7:18

Are you a gamer too? Casually, so I'm not a huge gamer. I do like video games, especially strategy ones. Um, I do enjoy that, but I'm not a huge gamer. I would say I would classify myself as a casual gamer. But you're absolutely right. You know, the whole thing of scientific compute on GPUs, which is you know the foundation of AI nowadays, right? Yeah, that originated from video games. Nvidia, remember Nvidia was originally a video game company, right? All the GPUs writing shaders and utilizing the hugely parallel architecture of GPUs, all of that was initially done for graphics shading. Yeah, and then from that it exploded obviously into general compute and later AI. But yes, all of those origins, origins of you know high performance compute, graphics compute, all of that comes from video games. So there is a big connection between AI and video games historically.

Roopinder 8:10

Are you by any chance uh going to be visiting with Nvidia while you're in the neighborhood?

Speaker 3 8:16

Shall we say um no, not this time. We we started talking to them about that more more in on Cambridge side of things. Um I mean, you know, NVIDIA is a a great company, like you know, as I said, because so many things have changed, like when you look at the history and and how their you know hardware has been applied to many different, quite different and often orthogonal applications is quite quite interesting.

Roopinder 8:37

We're we're gonna get to what Bonniebo does, don't worry. I I have that, it's definitely on my agenda. I just have so after reading up your background and what you're doing, uh, I just had so much to talk to

Rare Earth Risk Meets Motor Design

Roopinder 8:50

you about. Uh, one other thing, though, that's very timely, I feel, for our discussion. So, our president, President Trump, is over there talking over in China talking to President Xi about uh many things. Uh uh trade, for example, but I expect one of the things he's gonna be talking about that Xi is gonna present for discussion is probably rare earth. Uh China has the monop not say monopoly, has a stranglehold or a grip on rare earth, right? Rare and rare earths are used for the magnets and electric motors. And I think Monumu, I'm gonna let you tell me about it. Monumu is a is an alternative, I think, to the type of magnets, motors that use rare earth magnets, correct?

Speaker 3 9:42

Right, so you must be following those that that absolutely so geopolitics for us, and I think in general for engineering, you know, in the in general, is very important because geopolitics can affect a lot of things. Um, you know, tariffs, material costs, all of those things directly impact uh how we think about engineering design. Uh, you know, a certain design of uh say an electric motor today might be the cheapest one, but you know, next month something happens, the the ratio of I don't know, the price of steel to the price of electric, you know, rare F metals changes, and then suddenly your design that today was optimal for cost, tomorrow is not, right? Um so yes, all of those things are very important, and yes, obviously we do follow them. And you're right that monumental can absolutely help because what we do is we uh basically optimize complex engineering systems and we can optimize them for whatever the customer wants them to be optimized. Often it is cost, often it is material usage, often it's energy efficiency, etc. So, for example, in the supply chain issue, if a given company wants to really reduce the reliance on rare earth metals, we can optimize their electric motor designs or powertrains to reduce the amount of magnets that are needed. Or alternatively, we can find other types of motors that don't require earth metals at all and make them as efficient as possible. Now, you know, physics is physics, and you you know, magic doesn't really exist. So, in terms of energy density, if you put permanent magnets in there, it's always gonna be better than without permanent magnets. But the goal is to reduce the amount of permanent magnets as much as possible and still maintain the performance levels of the motor that the customer requires. But yeah, maybe uh just on a very high level, maybe I should just yes, please.

Monumo And System-Level Optimization

Roopinder 11:34

Let's see, we jumped on the deep end there. We jumped into a very specific application, right? But tell me more about tell me the monument, what what exactly is trying to solve and what it does, and and then your role too. You went from CTO to CEO, right?

Speaker 3 11:49

Correct, yes, that is correct. Um right, yes. So Monimo, I mean, maybe I will give you a very brief overview. So uh we're about four and a half years old, based in Cambridge, UK, although we have another office uh in Coventry, UK, which is another town in the UK, where we have a workshop and we can sort of look at some uh physical prototypes and some physical measurements, which is quite important because when we talk to engineering customers, an ability to sort of to work with actual hardware is often seen uh very valuable, you know, highly valuable. But the main premise is that we believe that uh when we think about engineering, we need to really think about it in a holistic and system-level way. And traditionally, a lot of engineering is done on a per component basis, and there are many reasons for that. There are reasons uh that are technical and there are reasons that are organizational for that. It's very natural for humans. Uh, if you have a complex problem uh that you're struggling to solve, you tend to break it down into smaller components and then look on work on those components in isolation, right? It's a natural thing to do, right? And obviously, that's how engineering has been done. You know, it's very difficult if you're an engineer to think about the design of a whole car, right? That's there's so many things to consider. So obviously, we break it down. You know, we have different engineers who will design the chassis, different engineers who will design the powertrain, different engineers who design, you know, suspension, etc. But even within those bigger components, you still have a lot of specialization. So if you look at, I don't know, say powertrain, you will have people who design the actual electric machine, you will have people who will work on the inverter, you will have people who work on the gearbox, etc.

Roopinder 13:28

It's a system of systems, right? You can't just be you can't uh I have to interject with a common perception of AI right now. Common the person on the street, they ask about AI, they say they think of questions like uh, hey, AI, make me a building, or AI, make me a car, and you know, and engineers know. It's just you have to do it a little bit. It's not how it works. This is systems, they're systems, you have to optimize each system, you have to develop each system. You know, the fact that uh you can say, Hey, AI, I wonder this myself, hey AI, help me electrify this car, right? What could be simpler? You've done it like a uh automotive company may have done it already several times, right? All it needs is take out the motor, take out the IC engine and put in the electric motor, right? Use that, make the space for the battery. It it should be very simple, but it's not like that if you try to do it because the moment you start modifying, let's say you want to make space for the battery, right?

Speaker 3 14:30

Then suddenly things move, your seats will move, your the suspension needs to change, and then suddenly that simple change that you just said suddenly spills out and affects literally everything about the design of the car, right? Um, so basically the system level view and system level optimization is very difficult, right? Because it has so many dependencies and so many parameters. And this is basically where our technology comes in. So, Monimo's premise is that we deploy deep tech, machine learning, AI, physics, AI models, etc., to basically solve this complex optimization problem at the system level. And we started by applying this to electric machines, so the powertrains, uh, both in generation and in motoring space. But uh the goal is to apply the same approach to any complex engineering systems. You know, by doing this system level optimization, that is exactly where we can handsafe, say, rare earth metals to customers, right? Because by looking at the system level, we can find trade-offs between the motor design, the inverter design, potential gear ratios, et cetera, that allow you to achieve the same performance of an electric motor in terms of torque, losses, etc., but with significantly fewer permanent magnets, for example, if that's what the customer

Motors Everywhere From Robots To Grids

Speaker 3 15:53

wants.

Roopinder 15:53

Okay, so Monibus's first specialization was really electric motors, correct? And now it's is it now have you already conquered that territory, as it were, and now you're expanding to other uh other systems, let's say in not just in vehicles, but in aircraft and wherever motors are needed.

Speaker 3 16:13

Yeah, so you know, first uh first thing to note really is that electric motors are extremely common, they're everywhere. People often think about oh, electric motors, you must be working with EVs. Yeah, and obviously we do work with EVs, but EVs is just one of the applications for electric motors. Electric motors are in washing machines, industrial pumps, car tools, right?

Roopinder 16:35

All the time I'm thinking about, you know, I do a lot of not a lot, but I work at what my favorite tools are the my Dewalt electric drill. And I see every model is getting smaller and smaller, exactly more compact, and though it must be the motor, right?

Speaker 3 16:50

Yeah, exactly, right? You know, the motor, the probably the gearing, the integration of all of that. So basically, electric motors are in a lot of different verticals with a lot of different applications. Um, so in a way, our first target is all of the electric motors for washing machines, drones, robotics. We should definitely talk a little bit more about robotics because that's a very interesting.

Roopinder 17:12

How many how many motors are in a robot?

Speaker 3 17:14

Like a lot, especially the humanoid ones that you know they have so many actuators. It's um when I first learned that I was very surprised. More than 50% of the bomb cost of a humanoid robot goes into actuators. Ah very high number. Yeah. So any improvements that you can make to make those actuators cheaper, better integrated with the rest of the uh the robotic design, you know, the savings are really, really large. So robotics is definitely a very interesting market for us. So is drones and the defense angle, obviously, as well.

Roopinder 17:50

Right. Are are most uh okay, you hear about drones a lot now with the military, with Ukraine, with the potential conflict with uh China and Taiwan, it's all gonna be drones, drone, drones. Uh are they electric? Are most of them electric motor?

Speaker 3 18:09

Um depends. So the really long-range ones are usually not, um, but the shorter ones, the say the interceptor drones or so, they're almost always electric. And in a way, for us, that's the the the nicest market to also work because you know if something is a clearly defensive, defensive, like an interceptor drone, then in a way it's easier to work with purely defensive applications that are not offensive.

Roopinder 18:34

Okay, I didn't realize so much of the cost with an Alubinai robot was was motors. I I had the perception it was going to be the computer, computers in in a road, right?

Speaker 3 18:44

No, I mean what I'm talking about is the bomb cost, obviously. There are a lot then other costs in terms of you know actual development of algorithms and controls and all of that, but in terms of bomb cost, um yeah, that's it's pretty high.

Roopinder 18:56

Yeah, yeah. Yeah, uh I read on your site uh that 50% of all electric electrical energy is used by motors. Yeah, it's crazy, right?

Speaker 3 19:08

Yeah, yeah. Electric motors are really a huge global market. Um, so you know, the basically electricity is consumed by electric motors, heating systems, that's the other big one. Uh all the key pods and all the heating applications. Um, and then obviously, you know, semiconductor uh application, you know, all of the electronics and all of that. But electric motors is yeah, it's about 50% of global electricity goes goes into electric motors. Even more crazy number that you know, that's the number that people really don't think about is uh that 80% of global electricity is generated with generators. Generated by electric motors. No, well no, generated by turbines and generators. Turbines, okay. Because the only source of power at scale that we have as humanity right now that does not require a turbine of some sort is solar. Everything else, even nuclear power plant, still hydroelectric as turbines that are that are at the bottom of the waterfall. Yeah, absolutely. And turbines are obviously, you know, also electric machines. They're very physic, like from the physics point of view, they're very similar to electric motors. They have electrom, you know, they have magnets, they have coils. So basically, our technology can be applied to also optimizing the efficiency of turbines. So, in a way, it's a very beautiful story where what we can do is we can improve the efficiency of the generation of electricity, and we can also improve the efficiency of the consumption of electricity, which is a very nice end-to-end story.

Roopinder 20:35

Yeah, it's a definitely energy sustainability angle to your company. Yeah, absolutely.

Speaker 3 20:40

Yeah, you know, energy is super important now, like in the age of AI, the data centers. We like all of this, like we in a way the bottleneck now is energy, right? The bottleneck is not necessarily building more CPUs or more GPUs, it's energy. Obviously, energy is not just generation, it's also delivery. So the you know, the grids, the national grid needs to you know upgrade the infrastructure. There's obviously a lot of talk about building power plants close to the data centers. You know, there's a lot of things that is happening that are happening in the in the energy sector. But as a whole, energy at the moment is a bottleneck.

Roopinder 21:18

Right, right. Oh, there's uh yeah, but reading about that, uh, about the energy that the data big data centers require, and it's getting scary now. It's like whole towns may be robbed of their energy because of the data center. And even scarier than that, now they're bringing back nuclear energy. To me, I know most engineers don't feel this way, but I'm not a big fan of nuclear energy. I feel like the genie's out of the bottle with that. Uh, I don't think we should be going that way, but it's bringing that prospect back to energy because nuclear plants can provide a lot of energy, right?

Speaker 3 21:56

Clean, right? When everything goes well, it is clean. Um yes. So yeah, absolutely. So you know, in a way, anything that we can do to help with efficiency of electric systems, I think, is huge beneficial.

Motor Types And The Torque Tradeoff

Roopinder 22:09

Yeah. So what are the big drawbacks of the type of motor that uh you're that you would be championing compared to the permanent, is it called PM permanent motors? You have to remember you're talking to mechanical engineer, not electrical engineer, right?

Speaker 3 22:24

Yes, so um basically there are few main types of electric motors that people use. Um there is the permanent magnet motors, so the PM motors, and there are a few variations of those, whether the magnets are inside of the rotor or on the surface of the rotor. So there are variations, but they all require permanent magnets. Um so that's one type. The other type is what is called wound rotor motors. So that's basically instead of putting magnets into your rotor, you put electromagnets. So, you know, basically the same as you have on the stator, you have electric, you know, coils, you can do the same thing on the rotor. Now, the problem with that is that you then need to somehow supply electricity to a rotor that is spinning. So obviously, you know, it's doable. I mean, people do that all the time, but it adds a little bit of complexity and therefore cost to the uh to the design.

Speaker 23:14

Right.

Speaker 3 23:15

And then there are simpler uh designs, simpler types of motors that don't require neither electromagnets nor magnets on the rotor, and those are mostly switch reluctance motors and synchronous reluctance motors. And they basically have a solid steel in the rotor with some flux barriers cut out of it, and those flux barriers then uh interact with the magnetic field and make the rotor spin. Now, the problem is that if you don't have any permanent magnets there and if you don't provide electricity onto the rotor, then uh the sort of the torque density of those motors is significantly lower uh than the ones with permanent magnets. Okay.

Roopinder 23:54

All right. Uh no, there's other types of motors that are that are that don't need the rare earth magnets, right? They don't need rare earth minerals and magnets made from them. I did a little bit of research on this, and correct me wherever I'm wrong. Switch reluctance motors, SRMs, no magnets. No magnets, but the the big problem, and this is something I think your company is helping with, is that torque ripple, correct?

Speaker 3 24:25

Yes, that is absolutely correct. So tell me what that is. What is torque ripple? So torque ripple is imagine um as your rotor rotates, okay, amount of torque that is being generated is not constant. Depending on the angle of the rotor, in respect in relation to the stator, sometimes the magnetic field will generate a little bit more torque and sometimes a little bit less torque. So basically, as the rotor rotates, amount of torque generated is not a nice constant amount, but it goes up and down. Now, this difference in torque generation then makes basically like vibrations, you know, because it's not uh, you know, it starts to basically shake a little bit, and obviously that has uh negative impact both in terms of the actual noise. Uh, you know, those motors tend to be noisier, but also on durability, right? Because if your system vibrates a little bit, then it will not last as long as a nice, smoothly operating system.

Roopinder 25:25

I see.

Speaker 3 25:26

I see the interesting bit here is that uh you can try to correct for this, right? You can try to reduce the stork ripple. And that is indeed one of the projects that we did, which was very interesting.

Beating Torque Ripple With Co-Design

Speaker 3 25:38

Uh, we did a system level optimization. So we asked our system to co-optimize the control strategy. So the control strategy tells you what currents, electric currents, are applied to what coils in your motor at what precise point in time to make the rotor rotate. And that control strategy can be quite complex, together with free form basically. So a lot of freedom in terms of defining the geometrical shape of the rotor. So that then created an optimization problem with a lot of parameters, because the free form of the geometry that had a lot of parameters, plus the control strategy also had a lot of parameters. And in a way, this exposes the main challenge with system optimization, which is that if you combine multiple systems into one optimization, in this case, it was control strategy and the geometry of the motor, then the number of parameters that you now have to optimize increases a lot because you have all the parameters of all the components put together into one bag, right? Now, the trick is, and the the unfortunate, you know, way unfortunately that's how it works, but complexity of optimization problems increases not linearly with number of parameters, but more like exponentially. So if you double number of parameters that you can optimize, the complexity of the optimization doesn't go up twice, it goes up of often orders of magnitude to up.

Roopinder 27:06

Is this the m m by n or m to the nth power problem? The exponential.

Speaker 3 27:12

Yeah, it basically multiplies, right? Because now if you imagine that you have four parameters to explore all of the combinations, you need to explore all the four. Now, if you if I add another parameter, I have to explore all the previous one plus all the combinations of the new one that I've just added, right? So it just keeps on multiplying, basically. Um so this is the this is a real challenge, right? Because that means that if you want to optimize something at the system level and you want to uh have the parameters of the whole system in your optimization, then you have two main challenges. One is you need to be able to simulate the whole system, so you need to be able to simulate, let's say, the electric motor, all the multi-physics in the electric motor, so electromagnetism, mechanical, thermal properties, all of that. But then on top of that, you need to know also how the gearbox will impact that, how the inverter parameters will impact that, right? So, in a way, your simulation stack needs to be able to deal with the whole system. Um, but even if you have that, then now you have this problem of number of parameters, because now you have all the parameters of all the components together, and your optimization problem becomes extremely difficult to a point that normal optimization approaches with normal tools just can't handle it. And that is exactly where monuments uh technology comes in.

Simulation Stack AI And Manufacturing Limits

Speaker 3 28:27

We have developed quite a lot of um you know very proprietary ways of dealing with this high-dimensional optimization problem, and that includes anything from AI models that can speed up uh physics calculations all the way to like uh optimization intelligence, so AI algorithms that make the optimization process more efficient.

Roopinder 28:47

So is many mode essentially an optimizing program? Is it uh or is it our solution?

Speaker 3 28:56

It's it's uh what I would say that it's an end-to-end solution, right? So the optimization is part of it, it's a very important part of it, but it is a part of it, right? Uh the other parts of it is the simulation itself, uh, where we can simulate all of those components and compute the right metrics, and the physics uh AI prediction models is the other part of it. So there is yes, there is the low-level physics uh AI prediction models, the optimization intelligence, which is the optimization basically algorithms and approaches, and the simulation, uh the simulation piece. Now, the fact that we have this whole stack that we control is actually very important because it means that we can, for example, use our simulation to generate large volumes of data whenever we need to. And obviously, data is very important for machine learning and AI. It also means that when we talk to a given customer, we can incorporate whatever manufacturing constraints they have into our simulation pipeline. And this is very important because it's you know, when you work with AI or any advanced 3D technology, it's super critical that what comes out of it is actually useful to your customers, right? So if my AI system would produce a motor design that is super efficient but cannot be made with the manufacturing process that my customer has, then the value of the design to my customer is not that high, right? Because they just can't make it. Um, so the ability to actually put the manufacturing constraints that customers have into our optimization and simulation pipeline uh is very important to the customers.

Roopinder 30:29

I'm looking at one of your examples on your website, and it says there was uh okay, 22 million FEM solves, and that's a mind-boggling number. Two questions. What uh what are you using for your simul for your solver? It's I it's this must be your own code, not not like a okay, all right.

Speaker 3 30:52

Yes, we're using our own code, and that number is very very large because remember, to get a torque of an electric motor, you need to actually solve a lot of uh electromagnetic uh fem solves to just get a torque curve in the control, right? So to just evaluate one design, one variation of a design, you need to do quite a lot of different fem solves.

Roopinder 31:14

Okay, all right. The other question is that if that many uh smoosh, that many solver, that much solving, and what this is just one type of uh physics. What but the the even motor design is a multi-physics, absolutely? Do you have solvers for all the different physics?

Speaker 3 31:34

Yes, so electromagnetism is the main one, it's the one that takes the most time and requires the most effort. So we obviously started there. But as you said, electric motors are multi-physics problems. If you only focus on one of the physics and then you run an optimization on top of it, the optimization algorithm will exploit that one physics and break all the other constraints. So, for example, you know, to a very, very, very standard example for that is from an electromagnetic point of view, it is the most optimal thing to move the magnets as close to the air gap in the rotor in the motor as possible.

Roopinder 32:09

Okay.

Speaker 3 32:10

But obviously, if I do that without concern of any mechanical properties, then when I spin my motor, then the magnets are just gonna break through a very thin layer of steel and fly off and break and destroy my motor, right? Absolutely, absolutely, right? So basically, if you only fit uh focus on one type of physics, your optimization process is not gonna deliver anything useful. You need to really consider all the types of physics and all the manufacturing constraints at the same time in order to get an actual optimal solution that makes sense. So, yes, we do that, we do mechanical uh source, obviously, and we do thermal uh thermal as well, because thermal is another very important one. Remember, you know, once you increase the temperature of your magnets past the curie point, then they just break and your whole motor is, you know, you can throw it away. Um, so you need to make sure that thermal limits uh inside of an electric motor are always obeyed as well, and that's very important.

Roopinder 33:05

Okay. Oh now, okay, so my head is always elsewhere uh up in the sky because uh aerodynamics. I have an interest in aerodynamics, even though talent in that space, but I'm always visualizing airflow. Do any of your solvers optimize for airflow? Like do like CFD?

Speaker 3 33:26

Yeah, no, so at the moment we don't do CFD. We focus on mechanical, electromagnetic, and thermal. Although uh in the future we will, because you know the goal is to optimize more and more complex engineering systems, and obviously, flow is important part for a lot of them. Uh, it is on our roadmap.

Roopinder 33:45

Oh, okay, okay. All right, so customers right now, a company has has customers in the electric motors, tier one space?

Speaker 3 33:54

Yes, so we have we have customers, both OEMs and tier one space, uh, in electric motors uh across different sectors. Uh so there is some automotive, some white goods, uh, some other other applications as well. Can't give you names because they're all very strict with NDAs, but yes, it's basically electric motors across different sectors.

Roopinder 34:13

Okay, all right. So, yeah, my drive my wife drives a Tesla. I have to ask, what kind of car do you drive?

Speaker 3 34:20

Interesting. I actually was a very early adopter of a Nissan Leaf. So I got a Nissan Leaf a very long time ago, uh, and I still have it. It's very good for local commuting within Cambridge. Uh but the range was uh, you know, it's good for local commuting, but not for long-range uh journeys. So I also have a Tesla, yes. Okay, so I have a Tesla and a Nissan Leaf.

Roopinder 34:42

Okay, all right. And I think Nissan would let you know if that would allow you to say if they were a customer. I don't think Tesla ever would. Okay, all right. But I think I think so. Elon Musk has always promised a better motor for his vehicle, right? Is is he talking about the type of motor we're discussing?

Speaker 3 35:01

This uh yes, I think that um definitely if you look at some you know um uh materials that are coming out of Tesla, uh they definitely spend time on optimizing their designs and they also start to think more and more about the system level optimization. So uh I I'm I'm I'm sure that it is being worked in at Tesla.

Roopinder 35:22

Yeah, yeah, they must they must be studying I with all this talk about rare earth and uh how we need to. I I think there was a national referendum or initiative to make more rare earth in the United States, but I don't know what came of it. I don't think that's it's true.

Speaker 3 35:39

I mean, the rare earth of you know they're not that rare, you know, they're actually quite common, right? It's just that extracting them is a very sort of destructive process. So yeah, it's sort of yeah, it's the mining process that's difficult, uh and not very nice for the environment and the people around it.

Roopinder 36:00

Yeah, but but but business good business opportunity, right? Because if people are exploring alternatives, then here you are with uh we can analyze, right?

Speaker 3 36:12

You know, absolutely right. So we can do two things, right? If you uh still want to use permanent magnet motors, then that's fine. We can optimize those designs to make them more cost effective, use a little bit less magnets, make them more you know uh energy efficient, you know, and even half a percent of efficiency that you can gain in your motor obviously translates into the battery requirements that you might have. And that obviously has very you know has high value. So we can do all of those, and then if you want to explore alternatives, so so motor types that don't have permanent magnets, then again, absolutely, we can help with finding the optimal designs for those motor types as well. And we've actually seen that quite a lot. A lot of our customers, their primary motor at the moment is still permanent magnet-based one, but all of them have some form of a permanent magnet-free motor on the back burner that they are working on for the future. Okay, it's very common.

Partnerships Pricing And Delivering Value

Roopinder 37:06

Yeah, yeah, okay. About the business, tell me more about the business and the business model. You don't have an off-the-shelf product with a close consumer UI or an app you can download, correct? This is a this yours is more like a process, isn't it? It's more like uh okay. So you put people on our on the site, for example, and you have them implement the system. We can so we can.

Speaker 3 37:34

We can basically we work closely with the customers. We actually often talk refer to them as partners, not customers, because it's it's a process where we work with them. Uh, you're right that uh our engineers get involved with their engineers to to set everything everything up, but then yes, at the moment we do run all the optimizations and all the uh results are generated internally, and we just deliver the results back to the customer. Um now we are um sort of exploring uh with especially some key customers and like strategic ones, giving them more direct access to the tool as well. Uh because you know, engineers like to play.

unknown 38:11

Yeah.

Speaker 3 38:12

Um so yeah, no, totally relate, right? So uh we we will be uh basically giving more direct access to what we have as well. It's just I prefer to start without it because then I didn't have to worry too much about the UI interfaces and all of that, and I could focus on actual technology that does the heavy lifting. Um, but we recognize that yeah, engineers like to play and giving them some form of access to um you know the the capabilities that we have directly would be beneficial.

Roopinder 38:40

It does sound more like a research partnership, RD partnership than it does a product sale or a product licensing, right? So uh that how do you price that? Is it just perk per customer?

Speaker 3 38:52

Like how we we obviously, you know, all of the customers we have are really large companies, right? So you have to be pretty custom with uh with the way you approach them because they have very different sort of structures of working with startups and companies and all of that. So we are quite flexible, but basically the main goal here is by running the optimization, you you get a design that has benefits. The simpler one is the simplest one to sort of price in a way is the reduction in the cost of materials that are needed to build a motor. So once you know what is that reduction is percentage-wise, and you know what the cost of the motor is more or less, and what are the volumes that the motor that the customer wants to make of that motor, that gives you a sort of an estimate of what is the value that our optimization has brought to the customer, right? And then you can use that as a in a way baseline on figure out okay, so what should be more or less the payment for Monimo in relation to that? Obviously, for what we charge is only a small fraction of the benefits that we bring, right? That's obvious, so that everybody wins, but in a way, it is related to the benefits that we bring and the volumes of motors that the customers make.

EV Futures Bay Area Trips And Waymo

Roopinder 40:02

Do you see a future where EVs have a natural life? I always thought of the eventual solution being an energy cell, like a hydrogen cell instead of a motor. Do you do you have any? I know you're you're you're now living in this world with with motors.

Speaker 3 40:20

No, so you know, in my very good question, right? From Monmo point of view, it actually doesn't really matter that much, right? We don't really care that much about what is the source of electricity. We still want to optimize the electric machine itself. Uh so whether it's hydrogen or a battery, it sort of doesn't matter that much. We can actually even work with electric motors that are in a hybrid, right? That's also perfectly fine. Now, hydrogen obviously has problems with, you know, I'm not an expert on this, obviously, but you know, main mainly purity of this, right? To get the purity that is needed to get the the fuel cells that actually work is quite expensive. Um, so we will see how it goes. As a concept, it's obviously um nice because it gives you a huge range and it gives you huge uh benefit in terms of how fast you charge your car. I mean, fuel it, right? But the range of car of electric cars with batteries is increasing now, right? And the battery technology is getting better and better. The car that I have, you know, I get like you know, nicely, even without putting too much effort into you know, too much of an eco-driving, I get like 250 miles out of it. Perfectly fine. If there is a good enough charging infrastructure, and the UK has a pretty decent charging infrastructure, I basically don't have range anxiety anymore. I know that I can go anywhere, and then if I need to charge, I just charge. Um, so I obviously, yeah, as you said, we don't have a crystal ball. I don't know whether the battery technology will have a huge breakthrough or whether the fuel cells are gonna uh lift, you know, take off. I don't know, but I'm pretty sure that electrification is the future. You know, what is the source of electricity? We will see. But in terms of efficiency of converting that energy into mechanical motion, electric motors are just so much superior to the combustion engines, right?

Roopinder 42:12

Right, right. You always see there's a long future ahead for motors, right? So okay, I get it. Okay, so how you're in California, how often do you come to the Bay Area area? Usually a few times a year, probably, yeah. Uh three times a year?

Speaker 3 42:27

Okay, is that for customers or for funding or both? So it's sort of customers, and obviously uh talking to VCs as well. Uh, you know, Silicon Valley is still you know that place for for deep tech uh VCs, and we are a deep tech company, you know, rooted in physical AI and engineering AI. So it's definitely something that VCs around here are interested in.

Roopinder 42:48

So for the old automotive trade, it was always the center of the universe was Detroit, right? Is there a center of the universe for uh EVs?

Speaker 3 42:57

Well, I mean, I would argue that at the moment the center of the universe for EVs is China.

Roopinder 43:02

China, yes.

Speaker 3 43:03

Well, I mean, in the EV space, it is, um, you know, and Asia in general. I mean, obviously, Korea is doing pretty well as well.

Roopinder 43:09

But um what about the US? Is there what do your travels customer travels take you all over the US?

Speaker 3 43:15

Or is it not really? It's usually usually um either Boston area or San Francisco area.

Roopinder 43:21

Oh yes, the two tech hubs. Yeah, yeah. Oh, so make sure you trap go in the Waymo. You have to experience the uh oh, I did.

Speaker 3 43:29

I use Waymo all the time.

Roopinder 43:30

Yeah, yeah, yeah.

Speaker 3 43:32

Aren't they great? Yeah, it's quite interesting. The first time I used one, I was a bit uh concerned for like about 10 seconds, and it's funny how quickly the brain just sort of accepts that. Oh, it's fine, it just works. Like literally, after 10 seconds, I was just like, Yeah, it's fine, it works. It's preferable, isn't it?

Roopinder 43:50

I mean, to compare the Ubers or stuff. I would I'd rather be in an away. I think that's yeah, it's a better experience.

Speaker 3 43:56

It is, and it's so many of them. I'm I you know, like it's a lot of Waymo on the streets. Yeah. So um, you know, looking forward to to uh to Waymo expanding as well uh into into other texts.

Roopinder 44:06

You know what I'll miss, Jarek? It's like I'll missed finding out the person's life story, like we did here. Oh, in the taxi, yes. Like who are you gonna talk to? Like what I always find these people are from interesting places and they have interesting lives, uh uh and I talk to them about it, right? But you know, I lot I'm gonna lose that. I'm gonna miss that part of it.

Speaker 3 44:25

Yes, I mean it's true that we we're in a way as a society now having less and less human-to-human interactions, right?

Roopinder 44:32

Yeah, yeah.

Speaker 3 44:33

Which sometimes is a little

Closing Thoughts And Listener Outreach

Speaker 3 44:36

bit sad.

Roopinder 44:36

All right, very good. Well, thank you so much for telling me for educating me on and what uh what your company does, and I'm glad I met you. Uh good luck with it. I think you have a great cause, especially the sustainability angle. I love that part. I think that uh it's all too often now just uh submerged under other under uh issues. But uh we should be much more concerned about uh sustainability and climate than we are because of all these other issues. But I really appreciate your time. Jarek was great, great to meet you. Thank you very much.

Speaker 45:12

Thank you for having me. And see you around, yes. See you around, I hope so. All right, bye bye.

Roopinder 45:19

Thank you for listening to FoDES, the Future of Design and Engineering Software Show, brought to you by ENGtechnica. I hope you have learned of a new application or technology that will help you with your job. If you have an application you think would be of interest to other engineers, please let me know by emailing me at roopinder at engtechnica.com or message me on LinkedIn.