AI-Designed Materials to Cool and Decarbonize Data Centers with Orbital
Jonathan Godwin is co-founder and CEO of Orbital Materials, an AI-first materials-engineering start-up. The company open-sourced Orb, a state-of-the-art simulation model, and now designs bespoke porous materials—its first aimed at cooling data-centres while capturing CO₂ or water. Jonathan shares how his DeepMind background shaped Orbital’s “design-before-experiment” approach, why the team chose data-center sustainability as a beachhead market, and what it takes to build a vertically integrated, AI-native industrial company. The conversation explores the future of faster, cheaper R&D, the role of advanced materials in decarbonization, and the leap from software to physical products.
Episode recorded on April 30, 2025 (Published on May 27, 2025)
In this episode, we cover:
[02:12] Johnny’s path from DeepMind to materials start-up
[04:02] Trial-and-error vs AI-driven design shift
[06:40] University/industry dynamics in materials R&D
[10:17] Generative agent plus simulation for rapid discovery
[13:01] Mitigating hallucinations with virtual experiments
[18:18] Choosing a “hero” product and vertical integration
[25:43] Dual-use chiller for cooling and CO₂ or water capture
[32:26] Partnering on manufacturing to stay asset-light
[35:58] Building an AI-native industrial giant of the future
[36:51]: Orbital’s investors
-
Cody Simms (00:00):
Today on Inevitable, our guest is Jonathan Godwin, Co-Founder and CEO at Orbital Materials.
(00:07):
Orbital Materials is a materials engineering company that's designing and deploying physical products for advanced industries, starting with data centers. In particular, the company is part of an emerging category of what I've been thinking of as AI-first companies. They're developing physical materials, in this case, their first product is a carbon capture material that they believe is suited for the heat profile of data center use cases specifically, but they do so by leading with AI on the R&D side. Earlier this year, they open-sourced Orb, a state-the-art AI model for simulating advanced materials.
(00:47):
I believe there will be a whole class of AI-first companies that dramatically alter traditional R&D cycles, and this is what I was interested in speaking with Jonathan about. Jonathan comes to Orbital Materials with an AI orientation. He previously was a senior research engineer at DeepMind, an artificial intelligence research lab that is a subsidiary of Google's parent company, Alphabet.
(01:12):
From MCJ, I'm Cody Simms, and this is Inevitable.
(01:19):
Climate change is inevitable. It's already here, but so are the solutions shaping our future. Join us every week to learn from experts and entrepreneurs about the transition of energy and industry.
(01:39):
Jonathan, welcome to the show.
Jonathan Godwin (01:40):
Thanks for having me on.
Cody Simms (01:42):
I'm excited to learn from you today all about the world of advanced materials and how AI is accelerating how we as a society can discover new things, but also interested to hear how you as a founder have ultimately decided that not only have you built this discovery engine, but you already know how you want to productize some things coming out of that discovery engine. So, there's a lot to unpack here.
(02:08):
Before we do that, let's hear from you. Maybe give us a bit of a personal introduction.
Jonathan Godwin (02:12):
Yes. I'm Jonathan, Jonny for short. I'm the CEO and co-founder of Orbital Materials. My journey to get here started in AI research. I've always been fascinated by the idea that you should be able to predict the future based upon data and felt that if you could do that in any form of arbitrary way at arbitrary power, then you had an extraordinarily powerful and exciting technology on your hands.
(02:39):
And so I started learning about AI and ended up joining a company now well-known but not necessarily super well-known at the time called DeepMind, primarily working on AI for science and applying these technologies to predict incredibly hard things, things that were way too complicated for humans to understand or too computationally intensive for us to calculate with physics simulations. And just felt that over the time that I was there, which is about five years, that these technologies were just so incredibly powerful, but we were limited in some ways by staying in the computational world. We were evaluating them in a purely software sense, but there were a huge need for these technologies to be out there being used to make new scientific discoveries and ultimately those scientific discoveries having impact in the world.
(03:29):
And that sort of company didn't really exist, and that's the genesis of Orbital Materials. The scientific area that we're starting with there is advanced materials, so things like batteries, semiconductors, but for us, the physical products that underpin data centers and the AI revolution.
(03:45):
So it's been a long journey to get here. I'm not a material scientist by training, but I felt that the power of these tools was something that had to be applied to a problem of that scale, and so became incredibly fascinated by this area. And I've been doing this company now for two and a half years and making, I think, a lot of progress.
Cody Simms (04:02):
In your work at DeepMind, were you focused on the material space at all?
Jonathan Godwin (04:08):
Yeah. I mean, I started off in AI for medical imaging. I was one of the lead authors on Google's breast cancer product where we trained an AI to spot tumors in breast cancer scans better than expert doctors were. That was an incredible first thing for us to work on and was really well perceived around the world, but then moved into AI for fundamental sciences.
(04:32):
And so I ended up leading a team building large-scale machine learning models for material science, specifically at that time on catalysts, things for decarbonized industrial processes, but those same techniques can be used across a whole range of advanced materials. And so the final time of my time at DeepMind was spent really focused on this stuff, and I think some of those key insights from that time were the things that really gave me the belief that we could start productionizing this and bringing this into a more commercial setting.
Cody Simms (05:02):
How have advanced materials historically been discovered, designed, innovated upon?
Jonathan Godwin (05:10):
Yeah, so it's been a process of human chemical intuition, human chemical reasoning, and trial and error. And actually, some of the most remarkable materials that have ever been discovered have been purely through chance. People who discovered them were actually pursuing something a bit different, saw something unusual in one of their experiments that they weren't seeing, and decided to switch tack.
(05:33):
So really, the experimental discovery often comes before the theoretical understanding in the discovery of advanced materials, and that's really been the case for the past 100, 150 years. That experimental nouse and the actual dexterity, the physical mechanical ability to run these experiments well have been the cornerstone about how we discover new advanced materials. We still don't really understand a lot about batteries, for example. So it's really an experimental process and we want to be able to flip that on its head by starting with design and using these new tools that give us an insight that just hasn't been possible before.
Cody Simms (06:12):
And I think of a lot of advanced material research as being one of the areas where private industry and universities work quite well together. It feels like historically there's been significant amount of collaboration between the university research ecosystem and the corporate R&D department work around this. Is that an accurate way of thinking about it?
Jonathan Godwin (06:40):
Yeah, I think that's certainly been true. I think government has funded a lot of this R&D in universities, and then fundamental R&D, and then as the promising things come out of that foundational science, that's when companies have often come in a very positive, symbiotic way to bring some of that commercialization forward. That, in some ways, is a little different to pharma and biotech and so it's a slightly different model here, which I think is one of the interesting things about starting a business. You need to think a little bit differently about what commercialization development looks like.
Cody Simms (07:12):
And these often might be two or three-year commissioned studies where you have multiple postdocs, like you said, working on a grant program that might've been co-funded by the government and/or a corporation around an idea, and then ultimately, you generate some IP and the IP is licensed somewhere and maybe a corporation commercializes it or buys the IP outright. That's my understanding of the R&D pipeline. I don't know if that holds true in terms of what you've experienced.
Jonathan Godwin (07:41):
I think that's absolutely true. I think the only exception to this is going to be in areas of the semiconductor industry where a lot of the very advanced R&D is really kept proprietary and secret and the semiconductor materials, companies like Applied Materials, TSMC, Samsung, and many others, particularly in East Asia, they lead on semiconductor materials development in a far more corporate industrial development program. And I think when you roll back in the West to the 1960s, the birth of the semiconductor industry, that was in Bell Labs, that's always had a more industrial focus. I think the R&D within industrial processes, catalysis and things like that, has been a lot more like the outline that you put forward.
Cody Simms (08:28):
Yeah, and I would say much in the petrochemical world as well has had this duality to it of university and corporate.
(08:35):
So what happens then when all of a sudden we can design a new advanced material on a computer in a few weeks, a few months? I don't know what kind of timeframes we're talking here, but it is going to change that R&D pipeline, I would assume, and it feels like we're kind of at the now moment, right, where this is happening now?
Jonathan Godwin (08:53):
I think that's absolutely right. I think that that capability, I think, fundamentally changes the industry and it makes far more companies be able to do their own foundational research and their own foundational R&D, not rely so much upon that university sector. So, I think you'd see an absolute increase in the total number of dollars spent on materials development.
(09:13):
I also think it gives a route for startups where there just hasn't been one in the past 30, 40, 50 years. There was an explosion, I think, of new industrial companies in the '60s and '70s, kind of stopped around there. Whereas I think there's new capability to move quickly, create new types of organizations that are AI-first and can do R&D significantly better and faster than anybody else. It means that potentially you can get to market quicker, you can get to market on fewer dollars, you can get to market in a far compressed amount of time, potentially 10X-ing that amount of time, and then that then gives the sort of dynamics where you can see startups start to thrive as well.
(09:54):
So, I think you'll also see an explosion in the number of companies that are going to be able to be successful in this sector in a way that's been really challenging before for a whole host of reasons.
Cody Simms (10:03):
Describe what materials discovery looks like using an AI. What is the Orbital Materials process for this and how does it differ from what might have been a heavy, like you said, experimental lab-driven model?
Jonathan Godwin (10:18):
Yeah. So of course we've still got to have a lab because we've actually got to make these things and verify that our predictions work, and that forms a really important grounding to training data to help fine-tune our models. So we still do the experiments, the experiments aren't going anywhere, but I think the difference is whereas before you would just try and brute force the landscape of things that you thought might plausibly work to then moving to a design base, so saying, "I'm going to really radically reduce the number of experiments that I run, but those were experiments 10 or 100 times more likely to be successful because I've designed them through AI." And that AI process, that's the big methodological organizational shift in a company that is doing AI-first design of new advanced materials.
(11:03):
How does that really work? What is the AI actually doing? Orbital, we are a group of people who have a lot of expertise in AI across a lot of different areas of AI. We're well-known for something called Orb, which is our AI accelerated simulations model. It simulates quantum physics but very, very fast and at very large scales. So that means that we can have accuracy on some of the funkiest things that go on in advanced materials in simulation in a way that's never been seen before. The fidelity and accuracy of that is off the charts compared to what it would otherwise have been, and that means that you can start doing computer-aided design in the way that engineers have done computer-aided design for ships or airplanes over the past 30, 40 years. That simulation model allows us to do the similar sort of thing, but for advanced materials.
(11:52):
That's only aspect. So that is a really important tool that we at Orbital Materials are very well-known for developing, but internally, we combine this into a generative AI chemistry agent. That chemistry agent has access to our digital lab notebooks, access to all the information of the experiments that we've run in the past, it has access to literature search so it can search for prior auths in this area, draw inferences across multiple different academic papers in a way that would be really hard for a human to do, use that for hypothesis generation, and then it can use our simulation tools to run virtual experiments to say, "Okay, well, if I were going to do this experiment, instead of going straight to the lab, I can run a really high-fidelity virtual experiment using my AI accelerated simulation, get information."
(12:38):
I can run a million of those at a time, search all of the space that I think is physically plausible, and then I come up with hypothesis that's incredibly well-grounded through the prior experiments, through the literature, and through our virtual experiments so that when we go into the lab, we've got a really strong sense of what we're going to make is going to be successful. And I think that's the thing that really changes.
Cody Simms (13:01):
I'm trying to think through some of the challenges you might face, and in my own very basic usage of desktop AIs, I get very obvious hallucinations from things that I'm asking that are just based on basic internet search. And fine, I get it, and maybe it makes its way into something I'm doing and misinforms my way of thinking about something. You're talking about if some kind of hallucination makes its way into your research, it starts to impact a production process which has real-world physical costs and problems. How do you navigate that?
Jonathan Godwin (13:40):
I think that's a really good question and it's one of the reasons why we have developed other tools around the base generative AI system. So our AI simulation is not going to hallucinate in the same way that a language model or a generative AI tool like ChatGPT would. And so these two things work across each other. The simulation can do some things a ChatGPT-type system can't, but that ChatGPT system has great access to literature, great access to hypothesis generation. The combination of those things radically reduces the possibility of simulation.
(14:13):
And then finally, we still have, for a lot of our experiments, humans run those experiments. We don't have a fully automated wet lab, and so the human intuition then takes place, like, "Okay, here's the suggestion. Before I go spend the next few hours doing this, now I'm just going to check it. I'm just going to think, 'Does this make plausible sense? Do I believe that what the AI has come up with has a good chance of success?'" And if we can spot a hallucination there, then we won't do it. We'll ask for another candidate. We'll ask it to go and check.
Cody Simms (14:41):
What I'm hearing is an incredibly complex, multivariant A/B test framework, which is really what the promise of quantum computing is, but it sounds like you are trying to bring that to bear today in a world where you can test many things. You're not trying to generate something on the fly necessarily. You're trying to test how it would react to various stimulus.
Jonathan Godwin (15:01):
Yeah, so that's where I think the fact that we have multiple AI systems really helps. We have a system that generates, so the generative AI system, and then we have a system that verifies through virtual experiments, and we can run lots and lots of those virtual experiments. As you said, how is it going to run under 20 different pressures? How is it going to run under different temperatures? What's going to happen if it comes into contact with this molecule? Is a chemical reaction going to happen and what is that chemical reaction? We can answer those sorts of questions in a really robust way through AI accelerated simulation. That complements the generative AI tools and the generative AI capabilities.
(15:37):
I think this is speaking to how sometimes I get a question which is like, "What's your AI program? How is your AI program different?" And there are certain things that Orbital do, I think, better than anyone else in the world, but it's wrong to think about Orbital or perhaps other companies in this space as a one-shot, "This is the approach that's going to end up solving AI for science."
(15:58):
The systems that are coming out at the moment, and there are many ways in which AI can accelerate a scientific endeavor, I think large language models one, and we use all of those, and every part of the process of developing a new advanced material, we see applications of AI across all of that. We spend a lot of time thinking about discovery, but when you get into some formulation, "Okay, well, I've got the base material, but does that need to be in a beaded form? How am I going to formulate it, shape it so that it has the very best properties?" That's a really hard question. Traditionally, people have spent a huge amount of the R&D budget for a material on that question, and AI is going to solve that too. It's going to be slightly different from that simulation model. It's going to be something perhaps more like a large language model, lots of different things that can help us with that question, but AI is going to really accelerate that.
(16:50):
And then finally, when you get to actually making it, is AI going to accelerate other design of manufacturing plants? Yeah, absolutely. I don't think Orbital's ever going to have a plant of our own, but we're going to be able to use AI to help our partners who are making our materials improve their manufacturing capabilities, absolutely, so we can reduce the cost of those materials.
(17:08):
So I think AI, when I think about Orbital Materials and I say that AI is going to accelerate advanced materials discovery, I mean it's accelerating all of that, and we're just here and there's one thing here and that's incredible, but the ambition and the vision is so much wider and it's incredibly exciting what's going to happen to that entire capability.
Cody Simms (17:27):
So this, I think, is a good opportunity for me to then ask the question of, and yet it seems like you are not building this company to just be a discovery and materials R&D factory. You have decided you are going to create some product lines and actually bring physical product into markets, which we should talk all about, for sure, because they're highly relevant to many of the topics we discuss on the show.
(17:54):
But before we even get into what those products are, how did you decide to essentially verticalize? As the CEO and as the entrepreneur here, it sounds like you've created this incredible horizontal discovery platform that could be a high-margin software business all by itself in theory, and you've decided to also go build a hero product. Explain a bit more about how and why.
Jonathan Godwin (18:19):
I love that phrase, "A hero product." I think it's a nice way of encapsulating what this vertical integration strategy requires from a company. Maybe it doesn't have to be a single hero product, but it's certainly you got a couple of shots and one of those shots has got to hit. I think luckily for us, certainly the signs are, these anchor customers being some of the largest companies in the world, that our first shots have hit, but there was a period where that was a real scary decision for us to make. So, it's a good question.
(18:47):
For me, the reason that I've always done this is to get stuff into the real world, and yes, you can feel like you're doing that by providing software, but your real belief, it was our company and our team that really did that. That comes from doing it yourself. So there is a non-trivial part of this question, which is just this is what the founding team wanted to do. This is what got us out of bed in the morning and got us really excited.
(19:14):
There are good business reasons to do this as well. Maybe though we could start with the reasons why not because I think it's good to hear those because we heard them a lot. When we'd go out and raise money a couple of years ago, we heard a lot of this.
Yin Lu (19:27):
Hey, everyone. I'm Yin, a partner at MCJ, here to take a quick minute to tell you about the MCJ Collective Membership.
(19:34):
Globally, startups are rewriting industries to be cleaner, more profitable, and more secure. And at MCJ, we recognize that a rapidly changing business landscape requires a workforce that can adapt. MCJ Collective is a vetted member network for tech and industry leaders who are building, working for, or advising on solutions that can address the transition of energy and industry. MCJ Collective connects members with one another with MCJ's portfolio and our broader network. We do this through a powerful member hub, timely introductions, curated events, and a unique talent matchmaking system, and opportunities to learn from peers and podcast guests. We started in 2019 and have grown to thousands of members globally.
(20:21):
If you want to learn more, head over to mcj.vc and click the membership tab at the top. Thanks and enjoy the rest of the show.
Cody Simms (20:29):
By the way, before you go there, just the idea of putting something in the real world, that really resonates with me. Regular listeners to the show may know that my very first job out of college in the late '90s was kind of accidentally in the original dot-com boom. Worked for multiple years in an internet company that was great, it was a rocket ship, it had an IPO, and then two years later,, the company was gone because the dot-com bust followed the dot-com boom. And realizing that I spent my early 20s working on something that just flat out didn't even exist anymore was kind of a bummer feeling.
(21:01):
And what I'm hearing from you is, yeah, you can build really amazing software, but at the end of the day, is there something that is tangible to what you're doing? And so I get that just intrinsic desire to do that.
Jonathan Godwin (21:12):
Yeah. It's something that really, I think, unites all of the people in our company. We've got a lot of experience machine learning folks working with us and who clearly have a love, and I still have a very deep love of AI and I follow the academic literature very closely and still code pretty frequently, but a lot of what you do is pretty ephemeral. You develop something that hits the benchmark and that lasts for a couple of months, and then unless you're continuing to push that state of the art, no one touches your system ever again. This is true if you're working at OpenAI. No one's using ChatGPT-3.5 from two and a half years ago. A lot of what you do just gets replaced really quickly and has value in pushing the field forward, but doesn't have longevity most of the time.
Cody Simms (22:00):
I feel like that's a big reason why over the last five years we've seen so many very experienced technologists moving into climate tech because they realize, "Hey, I've spent 15 years building stuff that's on a quick replacement cycle and want to have more real-world impact."
(22:15):
Anyway, we could, I'm sure, talk on these theories for a long time, but I want to get back to your question. What did you learn that was the counter-thesis, I guess?
Jonathan Godwin (22:22):
Yeah, yeah. I guess the counter-thesis is you've got a horizontal platform. People use the open-source version. We keep the best performing versions internally for our own development, but we release an open-source version of some of our software as a contribution to the research community, and that is used for semiconductors by pretty big companies. It's used in catalysis for chemistry. It's used even in bio materials. So, that's an incredibly exciting breadth.
(22:48):
The question is now why... You can't go and develop a biomaterial, a drug, and a semiconductor as a company, but you can create value across those chains, so why don't you try and take that tiny slice across lots of things rather than go vertical? So, that's one reason why not to be developing something physical.
(23:06):
The other reason is that AI is really in our DNA, developing AI is hard. It's competitive. You need to focus. And maybe the bright idea is you've got capability behind really great AI people, just continue to do that and build a really great AI system. Adding on new capabilities to your company maybe is something that's going to be challenging. Why not just stick to what you're good at?
(23:26):
And then just lingering concerns about how do you manage to bring a product to market without being a manufacturer? Because I don't think we want to be manufacturers. We definitely don't want to be manufacturers. Some companies, especially in climate tech, discover new technology. They've got to become manufacturers in order to capture that value in order to bring that product to market in a physical way, and that ends up being pretty capital-intensive. We didn't want to go down that route.
(23:53):
So, those were the lingering concerns. Very good reasons to be wary about going and developing a new product. But the reasons we decided to do that is that when you think about how many really incredible opportunities are there in material science over the course of the past 30 years, there have been some, I think maybe solar panels would be a good idea, although there weren't so many successful solar panel companies. Semiconductors have been a good opportunity, but there aren't millions of things that happen on a macro scale that create new markets that could be dominated by materials.
(24:30):
We thought the data centers was one of those and offered a really unique point in time where we could create a new generational materials company in the truest form. And generational materials companies have been some of the most successful companies of all time. Intel, I know Intel is not so much of a success story these days, but my gosh, that's incredible what was able to be accomplished by Intel by being a vertically integrated materials company. Advanced materials won the world's Second World War for Allies, if you believe some of the stories.
(25:03):
So there was this opportunity here that we felt to create something that was orders of magnitude more ambitious than being a software layer used by a scientist and industrial R&D across a lot of applications that we just didn't care about and we didn't think ultimately could be very large opportunities. This felt like a unique once-in-a-generation opportunity for us if we could grasp it, and we felt that that was what venture capital, Orbital Materials, and our team really wanted to do. That's a large part of the reason.
Cody Simms (25:30):
So you identified the generational moments of the data center boom and identified that materials could play a key role there. Where did you go from there? How did you get to the current products that you're now building?
Jonathan Godwin (25:43):
It's a combination. I think one of the very most challenging things we have to do, and I think it's something that continues to be highest on my priority list, is what's at the intersection of required for data centers, required for our target customers, large market, technically tractable within the amount of money and the timeframe that we have, and just purely technically tractable, like is it even possible to design a material that has the properties that we look for, and then that we can bring to market in a way that is asset-light so that we're not building manufacturing plants?
(26:20):
And the intersection of that is pretty challenging to find, right? That is the Goldilocks zone of product selection because I think a lot of the time, even with one or the the other, it's a big market or something, but innovation is very hard in there. It's a small market, innovation is easy, all of these different things. So, we spent a lot of time really trying to understand that.
(26:39):
For us, we came to what could be seen, I think, as non-consensus view, which is that we were entering into a market that had a great deal of competition. Even if we had better technology, and we were pretty confident in a number of areas we could develop better technology, the fact that we were a startup and new to this space would make it really hard to break in. People would choose the thing that was less risky because the company had a long track record.
(27:06):
So, we wanted to go in a slightly left-field way and start with sustainability. Sustainability was not going to be an afterthought for our first product. It was going to be the reason people bought it, and we decided to do this because the major players who account for over 70% of the entire data center market, the hyperscalers, are also the companies that spend most on sustainability and care most about sustainability and push the sustainability requirements down upon the tenants, the co-location providers.
Cody Simms (27:34):
And have very aggressive net-zero goals, and yet set all of those before the data center boom was as it is now. And so I think Microsoft's chief sustainability officer recently said, "We still have a moonshot goal. The moon is getting farther away from us as we do it."
Jonathan Godwin (27:49):
Yeah, that's absolutely right.
Cody Simms (27:51):
They're doing more to combat their emissions, but the footprint itself is growing bigger than I think they anticipated it because of the data centers.
Jonathan Godwin (27:59):
Yeah, that's absolutely right. And we felt that that was an underserved part of the market that people, for whatever reason, in some ways didn't think of it as their responsibility to develop these products, whatever it would be. We knew that there was some demand for essentially what we call as our first product, which we call a Dual Use Chiller.
(28:17):
So a chiller is something that cools down data centers, like a refrigerator, you might think of it like that, but for a data center, like an air conditioning unit. It's a little different. That's the right way to think about it. So, that's its first use. The second use of the dual use is then it uses that waste heat for a purpose, and that could either be CO2 capture, carbon removal, or it could be water capture for atmospheric water capture and use that water as a way of reducing the water footprint or for cooling or for whatever many of uses for water within the data center.
Cody Simms (28:49):
I mean, that sounds a lot like what a metal organic framework would typically do, which can be used for heat extraction and can be used for carbon capture. It's actually, I think, used in both of those today.
Jonathan Godwin (29:01):
Yeah. These are definitely going to be porous materials. There are plenty of different alternatives to metal organic frameworks as well. I think metal organic frameworks is one materials category amongst many that can try and solve this problem. But yeah, they're porous materials. That's absolutely right. And we felt that we could design a material that would really be tailored for the characteristics of that heat that's coming out of that data center.
Cody Simms (29:26):
So it's specific to the exact temperature, the exact setting that a data center is operating in?
Jonathan Godwin (29:31):
That's absolutely right. And therefore, that would offer a really great opportunity for these hyperscalers because they were going to buy a chiller anyway, so they're already looking to buy this product, and because they were already going to spend that money, the incremental cost of adding on this additional functionality and the fact they're mainly powered off waste heat, that meant that the overall cost of getting that water in a sustainable way or capturing that CO2 was going to be a lot less.
(29:59):
So it's a win-win for these customers, and that was our entry point because we believe that once you get that buy-in and that reputation, then you can move laterally into more traditional data center products because that brand and that reputation is so important for this industry.
Cody Simms (30:15):
And so just to make sure I fully understand, so you'll do on-site cooling of the data center, or heat extraction probably more accurately, from the data center and then use, you said, that waste heat product to power a direct air capture engine?
Jonathan Godwin (30:31):
Yeah. The precise usage of that waste heat, we're not talking about right now because we'll be talking about that in a little bit more detail later on in the year, but that waste heat is an essential part of the functioning of our direct air capture system, absolutely.
Cody Simms (30:44):
Super interesting. But back to your capital-light notion, you then wouldn't actually build the mechanical mechanism to do all of this? You would essentially design and create the IP and then work with an engineering partner to build these?
Jonathan Godwin (30:57):
Yeah. I think something that's really important for us is that our internal facilities, we all make those pilots. We'll fabricate the materials for that pilot and we'll do the chemical engineering, the mechanical engineering to make a system, the first-of-a-kind system. In our case, it's not first-of-a-kind in a sense of a chemical plant. It's this first-of-a-kind in the sense of a shipping container. So a lot smaller in terms of capital intensity, well within the budget of a seed-backed VC company.
(31:24):
And we basically use that system to get the purchase order for a scale-out of that system. And that purchase order is then what's required, I think, to then get a really scaled-up manufacturing partner to come deliver. But our conversation so far, we can't say exactly who we're working with, but have indicated that with that, there's a huge amount of demand to supply those things. So lots of manufacturers like making things and want to make more stuff and being able to basically have a guaranteed purchase order is something that really unlocks our ability to get those manufacturers on board.
(31:57):
We would never do that manufacturing, but we do do that first-of-a-kind fabrication. We're shipping our first shipping container to a data center towards the end of this year. We'd love to have you come visit. I think that would be great.
Cody Simms (32:09):
And then in order to make the sale, now you're dealing though not only just with, "Oh, we've got to get help to get this thing manufactured," but you are also having to work with the data center to make sure they have some kind of, if you're doing direct air capture of CO2, now they need a pipeline for what they do with the CO2 that you capture. It creates multiple cascading integration questions, I suppose, that you have to work through with them. Is that right?
Jonathan Godwin (32:34):
Yeah, we absolutely do do that work to make sure that it's easy to purchase and work with Orbital. We are not going to source that equipment, but we'll help them think through what those solutions are.
Cody Simms (32:46):
The reason I'm asking is it's such an interesting challenge for you as the founding team to have decided to go all in on, "Yeah, we're going to build some of these products," because there's additional complexity downstream of you to work through as opposed to just, "Hey, we're a software platform."
Jonathan Godwin (33:00):
Yeah. I think a lot of people come to starting a company with a core set of competencies and skills. Sometimes people are a complete package in that founding team, got all the skills they'll ever need in order to be a really successful company. I think a lot of the time though, and in some of the most successful companies, you think about Jeff Bezos starting Amazon, "May be a software company, but instead, I've decided to become a warehousing company and a logistics company as well."
Cody Simms (33:26):
And a grocery store chain.
Jonathan Godwin (33:27):
Because of the scale of the ambition for what he wanted to achieve. And yeah, now, what, it's like a satellite company too?
Cody Simms (33:33):
Right, exactly.
Jonathan Godwin (33:34):
It's incredible. And so I think often there's a sense of here's where I think the opportunity is, the scale of this opportunity is, and it's two or three steps away from where I am now. Do I take the leap and go build that capability both as a founder to lead a team that stretches across things that I haven't done before and take the risk that it doesn't work out to achieve that ambition, or do I stay within what I know that I can currently do now?
Cody Simms (34:02):
And does the software platform remain an R&D platform that third parties can work with you to use, or is it now primarily an engine that your own team uses for materials discovery?
Jonathan Godwin (34:14):
Yeah, so we have never sought to commercialize our software platform. We have released a portion of it open-source as a contribution back to the open-source and research communities. It helps establish Orbital Materials's leadership in this space, and we're not going to commercialize everything that can be used in the open-source package. So we felt that that was a really great move for us for reasons that were slightly exogenous to our core commercial focus.
Cody Simms (34:42):
So you will be then, if I'm understanding, you are a materials development company that is going to discover, design, and build and market materials, and really your big advantage is that you, in theory, have the best underlying AI platform that enables you to do that faster, cheaper than anyone else.
Jonathan Godwin (35:03):
Yeah. I would have a correction, which is we're not going to be. We are a company that has discovered a material and have customers for that material's product, in fact, two of our products. And so we are that. But I think certainly two years ago when we said that to a bunch of people, they're like, "There's no chance." Yeah, we've done an incredible amount and that is through the use of AI. And I think that absolutely the thing that we are doing is not just the fact that we've got incredible AI technology, but we're designing a next-generation industrial company around AI.
Cody Simms (35:37):
What you just said though is so key, and I think for people listening, we hear about AI companies and you think of everything as a software platform, but no, what about new companies in every industry that do the thing that companies in that industry have done for a long time except they were built AI-first, and think how disruptive that is going to be to a traditional industrial company.
Jonathan Godwin (35:58):
Yeah, absolutely, because these companies have sat on their CapEx notes or just incumbency for so long that they are some of the most un-innovative, un-risk-taking, risk-averse companies in the world. And so being able to be a fundamentally different company that is able to far reduce speed, have a better quality product brought to market quicker at far less cost, and continue to operate and serve that product at a higher quality and a reduced cost, I think is game-changing. And that's what I really think about when you think about Orbital Materials, and underlying that has got to be exceptional software capabilities and AI capabilities and pushing the frontier of what all those things are. We do do that, but ultimately that's seen in just a far better-quality product at a better price point.
Cody Simms (36:43):
And Jonathan, you've capitalized the business to date through venture, I think, is that right?
Jonathan Godwin (36:48):
Yeah, that's right.
Cody Simms (36:49):
So I don't know if you've announced anything about funding to date.
Jonathan Godwin (36:51):
Yeah. So our most recent investor, we raised a Series A round at the end of 2022 from Radical Ventures, who have been a great firm, great support. We raised our seed round about a year before that right at this time when people were still pretty skeptical about AI, and so some two fairly high-risk-taking seed firms, one in the US, one in Berlin called Fly, and then the US, the seed firm was Compound. So all three of those have been really great partners as leads of our fundraising rounds. And then recently, NVIDIA made a investment in Orbital, which is at the end of last year.
Cody Simms (37:26):
Awesome. Congrats. Just a small AI company called NVIDIA.
Jonathan Godwin (37:31):
Yeah, I don't know if you've heard of NVIDIA, but they've got something to do with this AI business, so.
Cody Simms (37:36):
Amazing. Well, anything else that we should cover or that I should have asked that we haven't hit on?
Jonathan Godwin (37:42):
No, I don't think so. I think we've covered it all. One of our key things for people to get their head around with this company is that we've got this incredible AI talent, and most people think of us and expect us to be a software company, but we spend so much of our time in bringing things to market to capture value and create what we think the next generation of industrial companies should look like.
(38:00):
So I was really happy to be able to spend some time on the thinking on all of that because I would encourage more people to do this sort of thing because it's scary as hell, but it is doing something that's really outside of your comfort zone. But I think that's how any great technology business is built, and we need more businesses that are putting stuff out in the physical world, and perhaps a few fewer software companies. I say that as a software guy, so I'm allowed to say that.
Cody Simms (38:24):
Well, I certainly wish you the best as your materials continue to get adopted, and hopefully we'll see, I don't know, what would it be, dozens of Orbital Materials products in the market over the next five to 10 years? Would that be the idea?
Jonathan Godwin (38:40):
Yeah, I think maybe a few fewer than that. I think we're getting up to maybe six, seven, but each of those will be large businesses in their own right. So I think eventually, maybe in 10 years' time, we'll be talking about dozens. Yeah.
Cody Simms (38:53):
Cool. Well, Jonathan, thanks for taking the time. I learned a ton and appreciate you sharing it with us.
Jonathan Godwin (38:59):
Great. Thanks so much, Cody. See you later.
Cody Simms (39:02):
Inevitable is an MCJ podcast. At MCJ, we back founders driving the transition of energy and industry and solving the inevitable impacts of climate change. If you'd like to learn more about MCJ, visit us at mcj.vc, and subscribe to our weekly newsletter at newsletter.mcj.vc. Thanks and see you next episode.
