Matthias: good Andre. ⁓ Welcome our new podcast of Constructor Capital. Very proud have you today with you ⁓ on our Andre is the of G-Core, one of our portfolio companies. And I wanted to start our discussion with, you know, maybe you doing a bit of an ⁓ introduction, what G-Core you ⁓ what is the mission the company? You know, how do you work? Give us a bit of a sense. you know who you are.
Andre Reitenbach: Yeah, thank you very much for having me here today. And yeah, I'm happy to present a bit about the company. I mean, we are not a real startup G co-existence 2014 and our mission is to provide easy to use HCI secure solutions. And especially since the recent couple of years, we did a strong focus on AI. And that's why we are in this cash first constructor today because constructor with us has the same focus on doubling down on. on the journey and I believe it's a great place to be.
Matthias: Good, fantastic. So yeah, so you're not a real startup. You have significant revenues. You have many customers. You started, I guess, more as a CDN company. So what was really the inflection point when you moved to AI, which was also the moment when we invested in, I think it was the 70 million round. So tell me a bit about the journey and where you started and where you are today.
Andre Reitenbach: Yes, and then Yeah, we were and we are a CDN company, but we are a bit more. I would say that we are in a distribution business and we distribute not only content, which is a definition of CDN. We distribute everything which is related to the Internet, compute, AI, security, storage and then, yeah, as well. And the reason why we started to this business, because initially we are coming from entertainment business and entertainment problem, especially gaming problem was always about provide low latency secure capacity instantly at the edge. So this was the turning point, especially back then in 2014. This was not an easy part. were almost no players in this field. we combined the things not like everyone else having the distribution problem, mainly our CDN one. said, okay, compute is actually really important to bring really close to to the user and also it started also to be quite significant for security space as well. And around 20, we recognized that the next big thing going to be actually AI driven because we saw already first signs that OpenAI going to happen. And we saw already startups appear in this field. then initially, Nvidia was, of course, always around and we had a really long with Nvidia. Not around the AI, but more around GPU to providing GPU capacity for gaming at the edge because latency was important, but also quality of delivery ⁓ quite important. And then we recognize that AI pretty much can happen soon. And one of our angles in this field was sovereignty because we're European company. And we started to partner with a company called Graphcore. And Graphcore was a Bristol-based AI-first company building so-called EPUs to train large models. So we started to approach this direction pre-ChadGPT. And when the churchy pity happened, the whole world changed. And for us, it was already clear thing, and we were already invested in this field. yeah, was the right choice.
Matthias: Okay, wow, that's interesting. So you were one of the first customers of NVIDIA and when they moved to AI, you moved to AI and you followed them and you had a new offering and long legacy, obviously, with the gaming history that you had with them.
Andre Reitenbach: exactly. actually, of our customers initially were a distribution channel for Nvidia because every ⁓ single of us who is building entertainment stuff in Internet and through Internet ⁓ also demand on the other side, on the customer side to use those products. And for that, you need actually Nvidia products, ⁓ GPU and today as well, EI products.
Matthias: Got it. Yeah, I mean, now you're, you know, AI company, AI edge company. You have this long history with, you know, important vendors. Where do you see, you know, in this whole AI world at the moment, the biggest, you know, pain point that you're trying to solve for governments, for Fortune 500 companies? You know, what is it really that people need to figure out at this stage?
Andre Reitenbach: All this topic is a complex one and initially everyone was quite excited about AI Neo clouds. But technically, AI clouds in essence is still a cloud, ⁓ the workload is moving from CPU GPU based compute. And initially problem was to ⁓ find right amount of energy close to the use case and have the access to this compute because there was a huge demand which was bigger than the potential supply and access capital of course was an ⁓ important Now the market is shifting from pure training cases, mean training cases still ⁓ exist on the market but demand on inference and on use cases where you use AI to produce products and produce demand and produce an uncovered demand on the other side, ⁓ growing. ⁓ we see right now two cases appearing. One of them is really to build AI factories, talking machines to serve this demand. And the second one is as well ⁓ so-called private ⁓ solutions because sovereignty ⁓ trusted and private solutions ⁓ to be more important, especially to prevent data grab and not and to prevent on providers because most of those providers are sitting ⁓ in geographies and ⁓ we ⁓ several ⁓ recently it's really important for some players to ⁓ rely this infrastructure to have full control ⁓ and the data but also the compute next to them. ⁓
Matthias: Okay, okay, I see. If we talk a bit about the ⁓ and the market. I mean, obviously you can buy AI inference from many different players. You can go to AWS, you can go to Azure, you can go to some of the Neo clouds like Nebios. What is it that makes you different? How do you solve the customer problems that you just mentioned differently than the other players?
Andre Reitenbach: The market of AI is developing ⁓ quick and of course many products on its own you can it commoditized business, ultimately hosting ⁓ GPUs at data center or secrets of data. It's a unit economics game I would even say like a red ocean. The complexity on GPU side is probably comes on top that you need to ideally to have a full access to power supply data centers where you can put liquid cold equipment because the density of those GPUs is increasing. actually ⁓ USP if you can have the access to the power ⁓ access the capital ⁓ access to capable to scale those big clusters infinitely. then have all of this ⁓ core we this world would just building 200, 300, 400,000 deployments in one or couple data centers. And most of them are still focused and are staying around US market while Europe is by the capacity is lagging behind. So this first of all advantages because we are European player and we focusing on customers who demand sovereignty, European sovereignty and this is one of the aspects. combination ⁓ of distribution network, Global Edge, and ⁓ GPU, our existing products around security, compute, storage. streaming and the combination of the solution, which it makes actually added value solution at the edge. And I believe the future going to be way more complex and you cannot be one trick pony just delivering one aspect like big training clusters or big inference clusters. A combination of providing EI combined with compute, combined with storage, combined with delivery, combined with security. This is what makes us unique. And this is what made us unique also in other verticals because security on its own is also a narrow ⁓ ⁓ and ⁓ The combination of this, this is what makes us unique. And as well the capability to complex, private, hybrid solutions. This is what ⁓ we help us on the journey forward. And we have a lot of demand of those Neo clouds many of those companies managed to save a lot of... Capacity, they have the power, they have the data centers, they have the capital access, but what they cannot lift is time to market because they need all of this complex software ⁓ experience of the last 10 years to address this market and ⁓ and cover the demand of the customers. So they might not only ⁓ but also full solution. And ⁓ you cannot build in one or two years. This is you need experience and full stack engineering ⁓ on the support side.
Matthias: Got it. So there's a strong technical aspect, the full solution that you have. The other thing that you mentioned, and I'd like to go a bit deeper there is you said, you you're a European company. Obviously, most AI companies are American companies. So is there a regulatory... tailwind that you're getting from being European? Do European customers want to buy AI cloud solutions? Or ⁓ they care at all? Or is it just different things that they care about?
Andre Reitenbach: We actually see demand specifically from this angle and you can speak about so-called air factories, European air factories and... Serenity aspect or trust aspect is quite important especially because every region has their own regulatory and that's the demand especially in Europe to have a fully domesticated European provider soaring provider also capability to build the sovereign fully controlled infrastructures in Europe, but not only in Europe also in Asia and other regions and We see demand coming not like in the past people spoke about Serenity, but it was not really something must have. The regulatory in Europe is right now really focusing to make sure that only companies based in Europe can be eligible to address those markets. And we right now participate in several projects for air factories, like the Finnish one, the Spanish one, the Italian one. hopefully can be, as a part of consortium, ⁓ we stay relevant in these aspects.
Matthias: Got it. Okay, so it's really a geographic play as well. thing that we ⁓ also attracted by is the fact that you have ⁓ than 200 points of presence. worldwide, ⁓ obviously historically important because you come from the gaming industry, CDN industry, low latency is a key point there. How important it for your new offerings now? Is that a real differentiator, the points of presence, the latency, and ⁓ maybe for which cases is most relevant?
Andre Reitenbach: With this aspect, we were quite early to the game ⁓ those use cases, when we started to pitch this story to our customers, most them, are just still ⁓ in training ⁓ time the inference time where the latency was not that big topic at that time. So we to focus on the journey of our customers from the very beginning, also building training classes, inference classes, the smaller scale European ⁓ aspect. But as it evolves... there are more more use cases appearing where we really understand where the demand going to come from and we strongly believe edge aspect, low latency aspect, bandwidth aspect will play a bigger role. First of all, just because again, there are regulatory perspectives where you need to bring those workloads and encapsulate the tokens or generation of the AI in some particular geography, having the latency expectation. And on top of that, we believe that the future going to be intelligent grid. It's going to be agentic AI. demand will come from agents performing tasks as close as possible to the use case. sometimes in hybrid and private environments, there will be physical AI, which is starting to be a bigger topic more and more. And especially if you have like C cable cuts and cable cuts, they need to make sure that its infrastructure is distributed and stable in case of breaches. you can serve it from the next available spot. So believe this agenti cloud in the future world of ⁓ omniverse the future world of agenti cloud, agenti AI, in the future world of physical AI, really will be like an energetic grid, electricity grid. It's a, I-grid will be, has the same importance of being resilient and then globally distributed. And so we're all going to move in this direction. And we see already more and more demands in this direction. And sometimes you can also follow from cost perspective, where's the availability now in some regions already night and some regions already day. So you can shift the workloads where it's available, probably more cost efficient, and you can work with this. bit.
Matthias: Got it. OK. So it's a real full package around the globe. If you talk about competitors, mean, obviously, as you said, it might even be a Red Ocean. There a lot of competitors. If you take traditional CDN business, obviously Akamai is dominant. If you take Cloud, you have much bigger players. Core Weave, you mentioned. Nibius. Can you frame in simple terms what it what is ⁓ about you and how that unique positioning is?
Andre Reitenbach: Yes, definitely. And there's always a question if it's possible at all to compete against hyperscalers. Of course, by the nature of the hyperscalers, have infinite resources. At the same time, we see right now happening that they are focusing so much on AI that we see disruptions and we see downtimes and less focus on general business and in impact services. We had so many downtimes from big players and is first of all the company started to look as well ⁓ use multi-CDN, multi-cloud, multi-year approach and then put in on several legs. They recognized that second players ⁓ or players, specialized players They are in many things as good as hyperscalers, but they are more flexible. They are ready to be more client oriented because they're big players. They're focusing mostly on top 14 guys on really big tickets and don't neglect sometimes the smaller tickets, which is big enough for companies like G-Core. We have way more flexibility in a combination and we are ready to adopt things to the workflows of the customers. help them also on hybrid and private journey. but hyperscaler sometimes doesn't do. And of course, most of the names you mentioned, are still US space. So there's a lot of space for companies like G-Core to act. And especially look at the edge in combination with the eye. This is almost still a blue ocean. There are almost no players. We see right now some players like Akamai and traditional CDN players starting to move in this direction because edge and distribution. start to be more important in combination with regular and commoditized environments. G-Core ⁓ opportunity way earlier. So if we can execute it right, if we can move fast enough, if we can get more on ⁓ capital, we definitely be significant in this field.
Matthias: Yeah, no, and we see that. I mean, we're very happy with the progress that you made this year, and we're looking even more forward to going to happen next year. So if we move a bit from competitors and your defensibility, et cetera, to partnerships, talked Nvidia earlier on, but you have also many other partners, HPE, Nokia, Intel, Vsat. How important is that and what is the special thing about these partnerships and how do they help you succeed beyond just providing computing capacity or infrastructure or selling you stuff?
Andre Reitenbach: One of the strengths of GECO is agility. We are still small enough to adopt to the market quicker than the big players. We are present in almost all countries across the world, but our presence and visibility is limited to a couple of geographies, a couple of countries. But the solutions we provide are quite competitive and quite great and have a good market fit. So the good way for us to penetrate the market is through partnerships. And for us, partnerships with vendors, OEMs, is a channel for us. It enriches their portfolio. It creates a good mix of products together to address customer needs better. all of this HP and ⁓ you mentioned, ⁓ have a great ⁓ penetration and they're looking for opportunities to ⁓ ⁓ products into the markets as well as capable to move quick enough. ⁓ So using ⁓ us additional value to sell their products better. And for us, of course, it's great because we create a sticky business for them.
Matthias: Got it, And I saw earlier this year, I saw you with Jensen on a photo when you visited him. So it seems like they really care about you and about what G-Core is doing.
Andre Reitenbach: Yes, it's an interesting fact because course we are still ⁓ quite a small player in this field, but Nvidia recognize that our software solution is a great niche product and it helps them to sell faster. thing they don't like is that they sell GPUs to a player, the AI factory, and they struggle to adopt it. Just imagine you invest billions into this equipment ⁓ then you cannot put it immediately into production ⁓ because missing some software adoptions, automation adoptions, and your customers cannot use your products. ⁓ which delays so Nvidia sell quick enough. So we do right now, we build those every way I and the private air clouds where we give them part of our stack or full stack sometimes ⁓ so they can really be so-called the air factories and sell faster and then create value faster, shorten up the time to market. ⁓ And is great for Nvidia because then they can sell faster. Our customers sell faster. And of course, for us, it's a great channel to succeed.
Matthias: Got it. No, that's good. Let's shift maybe, you know, from present to a bit in the future. And, you know, you've been in this industry for many, many years, probably 15 years with G-Core, probably even more before that. And as you know, we are, as Constructor Capital, we are investing in lot of... future compute technologies. So we're investor in some quantum companies, quantum computing companies, we're investor in what? Fontonic computing companies that might change, you know, if they're successful, the world and might become an add on to traditional GPUs on some areas, even a replacement. You know, what's your view? You know, how long will GPU dominance continue? there be other stacks coming next to it? How you see the industry and technology evolving?
Andre Reitenbach: I mean, and we're following your investments closely we're also discussing ⁓ about potential opportunities to collaborate closer together. I believe photonic approach is a great one. And especially the where you sometimes in some countries, don't even have enough power and power density in data centers to put AI capacity, ⁓ GPU capacity in data center because it's just simply not available. grid or you sometimes cannot get 60, 50, 100 or megawatt of power in one data center. It's quite complicated but the demand for AI is permanently growing I believe photonic approach will enable edge even faster because through the photonics the power need per rack power need per space is very slower very lower ultimately it will not stop EI growing ultimately it will still create more more demand on on the AI because the scale of growth will outgrow any efficiency through Photonic, but it will of course create a push to be more efficient per token, more efficient per power, more efficient per kilowatt. So it's a great approach to put more compute for the same power. And I believe we are already really soon gonna hit the available power in the world. And you see already, it's like uranium is gonna be the next big thing because there's a limited amount of uranium in the world. and plants as a way to grow ⁓ energy to GPU demand. So Photonic is definitely a good direction ⁓ and soon as this product is going to be ready it will definitely ⁓ hit market and the market will receive it quite well. ⁓ So looking forward to it. ⁓ And quantum technology it's as a great technology. We ourselves, we see ourselves on the distribution layer so whatever technology comes and whatever demand gonna happen on the market if you see demand for ⁓ quantum-based compute at the edge we definitely gonna adopt it immediately.
Matthias: Okay, good. Well, we hope that will adopt some of our company's technologies. That would obviously be great. It's also our mission to connect our ⁓ various companies. Another question about the future, right? mean, if you go back 15 years ago, or for 15 years, you basically were bootstrapped, then you did one larger investment round where we joined 70 million. How do you see the financial profile of G-Core evolving? So do you need amounts of money like ⁓ other, you Coreweave or AI companies will need, do you see you more on a lean path where you grow with the customer base? What's your vision there, the financial profile, if you want?
Andre Reitenbach: Yeah, it's a great question. mean, G-Core bootstrapped the business for the most of its time. So since 2014 until our first round, 10 years of our business were mainly bootstrap. And the reason for that was that if you don't take the GPU... the demand for money was quite controllable. Just put in comparison, a regular CPU server costs you around, depending on the workloads, 2 10,000 euros or dollars. And one server for GPU and AI workloads costs you 3 to 400,000 dollars. the order of magnitude is so far bigger that you cannot do it out of your cash flow. You cannot do it really through regular banking. You need to go to investors. to do the PE and then companies like Constructor just to lift this capacity. course, the risk profile is a different one because compute and the business we've been to, ⁓ was quite established and... it was quite easy to find a capital for this demand. GPU is entirely new business, the risk profiles were higher and nobody knows how it's going to evolve for the next 10 years. this was an important decision for us to go in this direction and raise capital to finance GPU capacity and AI capacity. As we quickly understood that being a European company brings also challenges because capital to finance GPU capacity in Europe is very limited comparing to the US market, we can find way more risk capital for more interesting terms. So our decision was that we focus, of course, to build GPU capacity and we are still looking for projects where we can do and find bankable clients for three years and more. But beside of that, decided that we're going to focus to partner with people who acquire GPUs, who have the capital to live thousands and thousands of GPUs. And we combine our efforts together to create value for customers and again, yeah factories ⁓ installations this is where customer invest in GPU ⁓ we do the rest where we can bring most its value so this is ⁓ we will be the biggest value we can bring to the table. And on top of that, we see a cross adoption across our existing businesses. And just to speak about security, I believe security going to be by far the biggest ⁓ evolving market where you need to withstand ⁓ ⁓ And using our in security industry and combining with EI, ⁓ going to a lot of value for our customers and where we can also generate quite good margins.
Matthias: Got it. Okay, so this is how it became more capital intense. mentioned before, right, you're a European company, ⁓ of the few players, capitalism more constrained. Do you see, ⁓ mean, there's a lot of discussions right now, ⁓ know, with the US trying to become more independent, basically giving up the old continent, seeing them as outdated. Do you see waking up ⁓ terms of ⁓ embracing AI revolution? ⁓ Do see that ⁓ their is not words, but they're putting some muscle behind it? How do you look at the landscape that we face today?
Andre Reitenbach: That's a really good question and we had just recently a panel together with Nebius and there's several observations that we see. First of all, you started to take GPU, EI business more seriously there's efforts right now, also European funded, ⁓ to the EI factories. Challenge what we see here is that it's ⁓ efforts build the supply. But we will see right now our gap is on the demand side, made in Europe and out of Europe. And if we don't solve... demand question in Europe, the question is why we will need to invest more into the supplier. So really, we also need, we hope that Europe will also invest more into the demand side, funding and helping to create more AI driven companies, companies who actually demand more compute, demand more European compute, demand more GPU, demand more AI made in Europe, based in Europe, and that Europe makes efforts to create and maintain and keep more startups and an industry in Europe, then ⁓ supply question is to solve on its own. Because ⁓ you just fund the air factories but there's no one who wants to use it, then we have a big challenge.
Matthias: Okay, okay, I see. So we need more Black Forest Lab type of companies in Europe too.
Andre Reitenbach: Exactly, exactly. And once they are appearing, we need to make sure that stay in Europe. And then it comes to the question that funding is one of the biggest aspects keep them here because most of the companies like to stay in Europe. But if they cannot find enough funding, then they just go elsewhere where they can find ⁓
Matthias: Okay. So is that also the biggest, mean, growth is very nice. I mean, we don't reveal numbers here in this forecast, but you know, we have seen a very nice trajectory over the last 24 months, I would say. So what is your biggest nightmare when you wake up? What could derail G-Cost growth trajectory? Is it the European demand? Is it something else? Is it threats? How do you look at that?
Andre Reitenbach: and exactly to your point. Companies like G-Core can only stay competitive if they have a competitive environment where you have startups, have the whole ecosystem where businesses and the environment require services like G-Core. if we don't have this healthy ecosystem top down, then of course ⁓ will be more challenging to compete with companies sitting in Asia, sitting in US where they have the ⁓ have the businesses who really demand the services and actually of demand from Asia and from US for services from G-Core and we would wish that this is happening in Europe and also grow out of Europe. Otherwise it's going to be challenging to be competitive over the term.
Matthias: Okay, so this is really also your biggest concern. Very good. Well, final question. mean, Andrea, that was very interesting and gives a great picture. what are the, to wrap it up, what are the things or the value propositions that give you really confidence that you will win in longer term, that you keep on the strike? What makes you optimistic? We talked about nightmare. What makes you optimistic? for the years to come.
Andre Reitenbach: What makes me optimistic is the huge demand of our services and we see finally because we we've been always known for being quite innovative, but it's also important that the innovation we create also hits the market well and there's actually a demand. ⁓ now we see a lot of demand through different channels or different services to G-Core ⁓ we see all of our hypothesis that the agentic AI inference at the edge, security are driven, it's happening. And on top of that, even our new experiments, especially with private software, with private AI, we hit the nerve of the market and we can manage to ⁓ and then get more financing for all those projects, we see ⁓ a great path because what we always did, we did products and we did business which was sustainable. We've been profitable. This is quite important. If you see many our competitions, they still do our bet where they ⁓ take over the ⁓ revenue ⁓ profitability. We've been in this space. We always build our business in a profitable way. ⁓ and this puts me in quite optimism. We've had a really great growth this year and we are already in vision even without additional funnel which can come in the Q2 and Q3 so we're gonna do really well next year and we are on the right path.
Matthias: Super amazing. Well, thank you, Andre, for joining me today in this podcast. I thought it was really insightful and we're very happy and looking forward to continuing the journey with you and G-Corp. Thank you.
Andre Reitenbach: Thank you very much. Thank you for time.
Andre Reitenbach: Yeah, thank you very much for having me here today. And yeah, I'm happy to present a bit about the company. I mean, we are not a real startup G co-existence 2014 and our mission is to provide easy to use HCI secure solutions. And especially since the recent couple of years, we did a strong focus on AI. And that's why we are in this cash first constructor today because constructor with us has the same focus on doubling down on. on the journey and I believe it's a great place to be.
Matthias: Good, fantastic. So yeah, so you're not a real startup. You have significant revenues. You have many customers. You started, I guess, more as a CDN company. So what was really the inflection point when you moved to AI, which was also the moment when we invested in, I think it was the 70 million round. So tell me a bit about the journey and where you started and where you are today.
Andre Reitenbach: Yes, and then Yeah, we were and we are a CDN company, but we are a bit more. I would say that we are in a distribution business and we distribute not only content, which is a definition of CDN. We distribute everything which is related to the Internet, compute, AI, security, storage and then, yeah, as well. And the reason why we started to this business, because initially we are coming from entertainment business and entertainment problem, especially gaming problem was always about provide low latency secure capacity instantly at the edge. So this was the turning point, especially back then in 2014. This was not an easy part. were almost no players in this field. we combined the things not like everyone else having the distribution problem, mainly our CDN one. said, okay, compute is actually really important to bring really close to to the user and also it started also to be quite significant for security space as well. And around 20, we recognized that the next big thing going to be actually AI driven because we saw already first signs that OpenAI going to happen. And we saw already startups appear in this field. then initially, Nvidia was, of course, always around and we had a really long with Nvidia. Not around the AI, but more around GPU to providing GPU capacity for gaming at the edge because latency was important, but also quality of delivery ⁓ quite important. And then we recognize that AI pretty much can happen soon. And one of our angles in this field was sovereignty because we're European company. And we started to partner with a company called Graphcore. And Graphcore was a Bristol-based AI-first company building so-called EPUs to train large models. So we started to approach this direction pre-ChadGPT. And when the churchy pity happened, the whole world changed. And for us, it was already clear thing, and we were already invested in this field. yeah, was the right choice.
Matthias: Okay, wow, that's interesting. So you were one of the first customers of NVIDIA and when they moved to AI, you moved to AI and you followed them and you had a new offering and long legacy, obviously, with the gaming history that you had with them.
Andre Reitenbach: exactly. actually, of our customers initially were a distribution channel for Nvidia because every ⁓ single of us who is building entertainment stuff in Internet and through Internet ⁓ also demand on the other side, on the customer side to use those products. And for that, you need actually Nvidia products, ⁓ GPU and today as well, EI products.
Matthias: Got it. Yeah, I mean, now you're, you know, AI company, AI edge company. You have this long history with, you know, important vendors. Where do you see, you know, in this whole AI world at the moment, the biggest, you know, pain point that you're trying to solve for governments, for Fortune 500 companies? You know, what is it really that people need to figure out at this stage?
Andre Reitenbach: All this topic is a complex one and initially everyone was quite excited about AI Neo clouds. But technically, AI clouds in essence is still a cloud, ⁓ the workload is moving from CPU GPU based compute. And initially problem was to ⁓ find right amount of energy close to the use case and have the access to this compute because there was a huge demand which was bigger than the potential supply and access capital of course was an ⁓ important Now the market is shifting from pure training cases, mean training cases still ⁓ exist on the market but demand on inference and on use cases where you use AI to produce products and produce demand and produce an uncovered demand on the other side, ⁓ growing. ⁓ we see right now two cases appearing. One of them is really to build AI factories, talking machines to serve this demand. And the second one is as well ⁓ so-called private ⁓ solutions because sovereignty ⁓ trusted and private solutions ⁓ to be more important, especially to prevent data grab and not and to prevent on providers because most of those providers are sitting ⁓ in geographies and ⁓ we ⁓ several ⁓ recently it's really important for some players to ⁓ rely this infrastructure to have full control ⁓ and the data but also the compute next to them. ⁓
Matthias: Okay, okay, I see. If we talk a bit about the ⁓ and the market. I mean, obviously you can buy AI inference from many different players. You can go to AWS, you can go to Azure, you can go to some of the Neo clouds like Nebios. What is it that makes you different? How do you solve the customer problems that you just mentioned differently than the other players?
Andre Reitenbach: The market of AI is developing ⁓ quick and of course many products on its own you can it commoditized business, ultimately hosting ⁓ GPUs at data center or secrets of data. It's a unit economics game I would even say like a red ocean. The complexity on GPU side is probably comes on top that you need to ideally to have a full access to power supply data centers where you can put liquid cold equipment because the density of those GPUs is increasing. actually ⁓ USP if you can have the access to the power ⁓ access the capital ⁓ access to capable to scale those big clusters infinitely. then have all of this ⁓ core we this world would just building 200, 300, 400,000 deployments in one or couple data centers. And most of them are still focused and are staying around US market while Europe is by the capacity is lagging behind. So this first of all advantages because we are European player and we focusing on customers who demand sovereignty, European sovereignty and this is one of the aspects. combination ⁓ of distribution network, Global Edge, and ⁓ GPU, our existing products around security, compute, storage. streaming and the combination of the solution, which it makes actually added value solution at the edge. And I believe the future going to be way more complex and you cannot be one trick pony just delivering one aspect like big training clusters or big inference clusters. A combination of providing EI combined with compute, combined with storage, combined with delivery, combined with security. This is what makes us unique. And this is what made us unique also in other verticals because security on its own is also a narrow ⁓ ⁓ and ⁓ The combination of this, this is what makes us unique. And as well the capability to complex, private, hybrid solutions. This is what ⁓ we help us on the journey forward. And we have a lot of demand of those Neo clouds many of those companies managed to save a lot of... Capacity, they have the power, they have the data centers, they have the capital access, but what they cannot lift is time to market because they need all of this complex software ⁓ experience of the last 10 years to address this market and ⁓ and cover the demand of the customers. So they might not only ⁓ but also full solution. And ⁓ you cannot build in one or two years. This is you need experience and full stack engineering ⁓ on the support side.
Matthias: Got it. So there's a strong technical aspect, the full solution that you have. The other thing that you mentioned, and I'd like to go a bit deeper there is you said, you you're a European company. Obviously, most AI companies are American companies. So is there a regulatory... tailwind that you're getting from being European? Do European customers want to buy AI cloud solutions? Or ⁓ they care at all? Or is it just different things that they care about?
Andre Reitenbach: We actually see demand specifically from this angle and you can speak about so-called air factories, European air factories and... Serenity aspect or trust aspect is quite important especially because every region has their own regulatory and that's the demand especially in Europe to have a fully domesticated European provider soaring provider also capability to build the sovereign fully controlled infrastructures in Europe, but not only in Europe also in Asia and other regions and We see demand coming not like in the past people spoke about Serenity, but it was not really something must have. The regulatory in Europe is right now really focusing to make sure that only companies based in Europe can be eligible to address those markets. And we right now participate in several projects for air factories, like the Finnish one, the Spanish one, the Italian one. hopefully can be, as a part of consortium, ⁓ we stay relevant in these aspects.
Matthias: Got it. Okay, so it's really a geographic play as well. thing that we ⁓ also attracted by is the fact that you have ⁓ than 200 points of presence. worldwide, ⁓ obviously historically important because you come from the gaming industry, CDN industry, low latency is a key point there. How important it for your new offerings now? Is that a real differentiator, the points of presence, the latency, and ⁓ maybe for which cases is most relevant?
Andre Reitenbach: With this aspect, we were quite early to the game ⁓ those use cases, when we started to pitch this story to our customers, most them, are just still ⁓ in training ⁓ time the inference time where the latency was not that big topic at that time. So we to focus on the journey of our customers from the very beginning, also building training classes, inference classes, the smaller scale European ⁓ aspect. But as it evolves... there are more more use cases appearing where we really understand where the demand going to come from and we strongly believe edge aspect, low latency aspect, bandwidth aspect will play a bigger role. First of all, just because again, there are regulatory perspectives where you need to bring those workloads and encapsulate the tokens or generation of the AI in some particular geography, having the latency expectation. And on top of that, we believe that the future going to be intelligent grid. It's going to be agentic AI. demand will come from agents performing tasks as close as possible to the use case. sometimes in hybrid and private environments, there will be physical AI, which is starting to be a bigger topic more and more. And especially if you have like C cable cuts and cable cuts, they need to make sure that its infrastructure is distributed and stable in case of breaches. you can serve it from the next available spot. So believe this agenti cloud in the future world of ⁓ omniverse the future world of agenti cloud, agenti AI, in the future world of physical AI, really will be like an energetic grid, electricity grid. It's a, I-grid will be, has the same importance of being resilient and then globally distributed. And so we're all going to move in this direction. And we see already more and more demands in this direction. And sometimes you can also follow from cost perspective, where's the availability now in some regions already night and some regions already day. So you can shift the workloads where it's available, probably more cost efficient, and you can work with this. bit.
Matthias: Got it. OK. So it's a real full package around the globe. If you talk about competitors, mean, obviously, as you said, it might even be a Red Ocean. There a lot of competitors. If you take traditional CDN business, obviously Akamai is dominant. If you take Cloud, you have much bigger players. Core Weave, you mentioned. Nibius. Can you frame in simple terms what it what is ⁓ about you and how that unique positioning is?
Andre Reitenbach: Yes, definitely. And there's always a question if it's possible at all to compete against hyperscalers. Of course, by the nature of the hyperscalers, have infinite resources. At the same time, we see right now happening that they are focusing so much on AI that we see disruptions and we see downtimes and less focus on general business and in impact services. We had so many downtimes from big players and is first of all the company started to look as well ⁓ use multi-CDN, multi-cloud, multi-year approach and then put in on several legs. They recognized that second players ⁓ or players, specialized players They are in many things as good as hyperscalers, but they are more flexible. They are ready to be more client oriented because they're big players. They're focusing mostly on top 14 guys on really big tickets and don't neglect sometimes the smaller tickets, which is big enough for companies like G-Core. We have way more flexibility in a combination and we are ready to adopt things to the workflows of the customers. help them also on hybrid and private journey. but hyperscaler sometimes doesn't do. And of course, most of the names you mentioned, are still US space. So there's a lot of space for companies like G-Core to act. And especially look at the edge in combination with the eye. This is almost still a blue ocean. There are almost no players. We see right now some players like Akamai and traditional CDN players starting to move in this direction because edge and distribution. start to be more important in combination with regular and commoditized environments. G-Core ⁓ opportunity way earlier. So if we can execute it right, if we can move fast enough, if we can get more on ⁓ capital, we definitely be significant in this field.
Matthias: Yeah, no, and we see that. I mean, we're very happy with the progress that you made this year, and we're looking even more forward to going to happen next year. So if we move a bit from competitors and your defensibility, et cetera, to partnerships, talked Nvidia earlier on, but you have also many other partners, HPE, Nokia, Intel, Vsat. How important is that and what is the special thing about these partnerships and how do they help you succeed beyond just providing computing capacity or infrastructure or selling you stuff?
Andre Reitenbach: One of the strengths of GECO is agility. We are still small enough to adopt to the market quicker than the big players. We are present in almost all countries across the world, but our presence and visibility is limited to a couple of geographies, a couple of countries. But the solutions we provide are quite competitive and quite great and have a good market fit. So the good way for us to penetrate the market is through partnerships. And for us, partnerships with vendors, OEMs, is a channel for us. It enriches their portfolio. It creates a good mix of products together to address customer needs better. all of this HP and ⁓ you mentioned, ⁓ have a great ⁓ penetration and they're looking for opportunities to ⁓ ⁓ products into the markets as well as capable to move quick enough. ⁓ So using ⁓ us additional value to sell their products better. And for us, of course, it's great because we create a sticky business for them.
Matthias: Got it, And I saw earlier this year, I saw you with Jensen on a photo when you visited him. So it seems like they really care about you and about what G-Core is doing.
Andre Reitenbach: Yes, it's an interesting fact because course we are still ⁓ quite a small player in this field, but Nvidia recognize that our software solution is a great niche product and it helps them to sell faster. thing they don't like is that they sell GPUs to a player, the AI factory, and they struggle to adopt it. Just imagine you invest billions into this equipment ⁓ then you cannot put it immediately into production ⁓ because missing some software adoptions, automation adoptions, and your customers cannot use your products. ⁓ which delays so Nvidia sell quick enough. So we do right now, we build those every way I and the private air clouds where we give them part of our stack or full stack sometimes ⁓ so they can really be so-called the air factories and sell faster and then create value faster, shorten up the time to market. ⁓ And is great for Nvidia because then they can sell faster. Our customers sell faster. And of course, for us, it's a great channel to succeed.
Matthias: Got it. No, that's good. Let's shift maybe, you know, from present to a bit in the future. And, you know, you've been in this industry for many, many years, probably 15 years with G-Core, probably even more before that. And as you know, we are, as Constructor Capital, we are investing in lot of... future compute technologies. So we're investor in some quantum companies, quantum computing companies, we're investor in what? Fontonic computing companies that might change, you know, if they're successful, the world and might become an add on to traditional GPUs on some areas, even a replacement. You know, what's your view? You know, how long will GPU dominance continue? there be other stacks coming next to it? How you see the industry and technology evolving?
Andre Reitenbach: I mean, and we're following your investments closely we're also discussing ⁓ about potential opportunities to collaborate closer together. I believe photonic approach is a great one. And especially the where you sometimes in some countries, don't even have enough power and power density in data centers to put AI capacity, ⁓ GPU capacity in data center because it's just simply not available. grid or you sometimes cannot get 60, 50, 100 or megawatt of power in one data center. It's quite complicated but the demand for AI is permanently growing I believe photonic approach will enable edge even faster because through the photonics the power need per rack power need per space is very slower very lower ultimately it will not stop EI growing ultimately it will still create more more demand on on the AI because the scale of growth will outgrow any efficiency through Photonic, but it will of course create a push to be more efficient per token, more efficient per power, more efficient per kilowatt. So it's a great approach to put more compute for the same power. And I believe we are already really soon gonna hit the available power in the world. And you see already, it's like uranium is gonna be the next big thing because there's a limited amount of uranium in the world. and plants as a way to grow ⁓ energy to GPU demand. So Photonic is definitely a good direction ⁓ and soon as this product is going to be ready it will definitely ⁓ hit market and the market will receive it quite well. ⁓ So looking forward to it. ⁓ And quantum technology it's as a great technology. We ourselves, we see ourselves on the distribution layer so whatever technology comes and whatever demand gonna happen on the market if you see demand for ⁓ quantum-based compute at the edge we definitely gonna adopt it immediately.
Matthias: Okay, good. Well, we hope that will adopt some of our company's technologies. That would obviously be great. It's also our mission to connect our ⁓ various companies. Another question about the future, right? mean, if you go back 15 years ago, or for 15 years, you basically were bootstrapped, then you did one larger investment round where we joined 70 million. How do you see the financial profile of G-Core evolving? So do you need amounts of money like ⁓ other, you Coreweave or AI companies will need, do you see you more on a lean path where you grow with the customer base? What's your vision there, the financial profile, if you want?
Andre Reitenbach: Yeah, it's a great question. mean, G-Core bootstrapped the business for the most of its time. So since 2014 until our first round, 10 years of our business were mainly bootstrap. And the reason for that was that if you don't take the GPU... the demand for money was quite controllable. Just put in comparison, a regular CPU server costs you around, depending on the workloads, 2 10,000 euros or dollars. And one server for GPU and AI workloads costs you 3 to 400,000 dollars. the order of magnitude is so far bigger that you cannot do it out of your cash flow. You cannot do it really through regular banking. You need to go to investors. to do the PE and then companies like Constructor just to lift this capacity. course, the risk profile is a different one because compute and the business we've been to, ⁓ was quite established and... it was quite easy to find a capital for this demand. GPU is entirely new business, the risk profiles were higher and nobody knows how it's going to evolve for the next 10 years. this was an important decision for us to go in this direction and raise capital to finance GPU capacity and AI capacity. As we quickly understood that being a European company brings also challenges because capital to finance GPU capacity in Europe is very limited comparing to the US market, we can find way more risk capital for more interesting terms. So our decision was that we focus, of course, to build GPU capacity and we are still looking for projects where we can do and find bankable clients for three years and more. But beside of that, decided that we're going to focus to partner with people who acquire GPUs, who have the capital to live thousands and thousands of GPUs. And we combine our efforts together to create value for customers and again, yeah factories ⁓ installations this is where customer invest in GPU ⁓ we do the rest where we can bring most its value so this is ⁓ we will be the biggest value we can bring to the table. And on top of that, we see a cross adoption across our existing businesses. And just to speak about security, I believe security going to be by far the biggest ⁓ evolving market where you need to withstand ⁓ ⁓ And using our in security industry and combining with EI, ⁓ going to a lot of value for our customers and where we can also generate quite good margins.
Matthias: Got it. Okay, so this is how it became more capital intense. mentioned before, right, you're a European company, ⁓ of the few players, capitalism more constrained. Do you see, ⁓ mean, there's a lot of discussions right now, ⁓ know, with the US trying to become more independent, basically giving up the old continent, seeing them as outdated. Do you see waking up ⁓ terms of ⁓ embracing AI revolution? ⁓ Do see that ⁓ their is not words, but they're putting some muscle behind it? How do you look at the landscape that we face today?
Andre Reitenbach: That's a really good question and we had just recently a panel together with Nebius and there's several observations that we see. First of all, you started to take GPU, EI business more seriously there's efforts right now, also European funded, ⁓ to the EI factories. Challenge what we see here is that it's ⁓ efforts build the supply. But we will see right now our gap is on the demand side, made in Europe and out of Europe. And if we don't solve... demand question in Europe, the question is why we will need to invest more into the supplier. So really, we also need, we hope that Europe will also invest more into the demand side, funding and helping to create more AI driven companies, companies who actually demand more compute, demand more European compute, demand more GPU, demand more AI made in Europe, based in Europe, and that Europe makes efforts to create and maintain and keep more startups and an industry in Europe, then ⁓ supply question is to solve on its own. Because ⁓ you just fund the air factories but there's no one who wants to use it, then we have a big challenge.
Matthias: Okay, okay, I see. So we need more Black Forest Lab type of companies in Europe too.
Andre Reitenbach: Exactly, exactly. And once they are appearing, we need to make sure that stay in Europe. And then it comes to the question that funding is one of the biggest aspects keep them here because most of the companies like to stay in Europe. But if they cannot find enough funding, then they just go elsewhere where they can find ⁓
Matthias: Okay. So is that also the biggest, mean, growth is very nice. I mean, we don't reveal numbers here in this forecast, but you know, we have seen a very nice trajectory over the last 24 months, I would say. So what is your biggest nightmare when you wake up? What could derail G-Cost growth trajectory? Is it the European demand? Is it something else? Is it threats? How do you look at that?
Andre Reitenbach: and exactly to your point. Companies like G-Core can only stay competitive if they have a competitive environment where you have startups, have the whole ecosystem where businesses and the environment require services like G-Core. if we don't have this healthy ecosystem top down, then of course ⁓ will be more challenging to compete with companies sitting in Asia, sitting in US where they have the ⁓ have the businesses who really demand the services and actually of demand from Asia and from US for services from G-Core and we would wish that this is happening in Europe and also grow out of Europe. Otherwise it's going to be challenging to be competitive over the term.
Matthias: Okay, so this is really also your biggest concern. Very good. Well, final question. mean, Andrea, that was very interesting and gives a great picture. what are the, to wrap it up, what are the things or the value propositions that give you really confidence that you will win in longer term, that you keep on the strike? What makes you optimistic? We talked about nightmare. What makes you optimistic? for the years to come.
Andre Reitenbach: What makes me optimistic is the huge demand of our services and we see finally because we we've been always known for being quite innovative, but it's also important that the innovation we create also hits the market well and there's actually a demand. ⁓ now we see a lot of demand through different channels or different services to G-Core ⁓ we see all of our hypothesis that the agentic AI inference at the edge, security are driven, it's happening. And on top of that, even our new experiments, especially with private software, with private AI, we hit the nerve of the market and we can manage to ⁓ and then get more financing for all those projects, we see ⁓ a great path because what we always did, we did products and we did business which was sustainable. We've been profitable. This is quite important. If you see many our competitions, they still do our bet where they ⁓ take over the ⁓ revenue ⁓ profitability. We've been in this space. We always build our business in a profitable way. ⁓ and this puts me in quite optimism. We've had a really great growth this year and we are already in vision even without additional funnel which can come in the Q2 and Q3 so we're gonna do really well next year and we are on the right path.
Matthias: Super amazing. Well, thank you, Andre, for joining me today in this podcast. I thought it was really insightful and we're very happy and looking forward to continuing the journey with you and G-Corp. Thank you.
Andre Reitenbach: Thank you very much. Thank you for time.
