Europe's AI Infrastructure: Can Nscale Close the Gap to US Giants?
Source: Startup Insider – Original Episode (episode in German)
$2 billion – and the question of whether Europe is actually building its own AI infrastructure or just financing expensive hardware. Björn Rieckhoff discusses the Nscale Series C with Jan Thomas: Europe's largest AI infrastructure round to date. GPU assets as bankable securities, the role of debt financing, the blurring line between venture capital and infrastructure investment – and why the investor landscape for deals like this looks nothing like traditional tech VC.
Key themes in this episode
- Berlin auf die Eins: Community energy in the Berlin startup scene and the question of whether the competition between Berlin and Munich actually benefits Europe.
- Advisory OS and AI tools: Why Björn is cannibalizing his own advisory business with AI agents – and why value-based pricing is the future of consulting.
- Services as the new software: Sequoia's thesis that hyper-specialized services will produce trillion-dollar companies – and what that means for founders and investors.
- Nscale Series C: $2 billion for Europe's largest AI infrastructure round – GPU clusters, data centers, and the value chain behind AI compute.
- GPU assets as bankable securities: How physical infrastructure is shifting the financing logic from venture capital to debt-backed infrastructure projects.
- European tech sovereignty: Why Nscale as a European player is politically relevant – and whether an IPO is the most likely exit channel.
Transcript
This transcript has been edited for readability. The content and statements have not been changed. The original conversation was in German; this is an English translation.
Jan Thomas: Björn, hi, great to see you!
Björn Rieckhoff: Hi!
Jan Thomas: Great that we're talking. Back on the podcast again. We just saw each other. That was really nice. We should probably talk about that briefly. Berlin auf die Eins. What feeling did you go home with?
Björn Rieckhoff: A very positive one. I think it's great that there's a real community movement that clearly strikes a nerve, when I talk to Berlin founders and investors. There was just a great energy in the room and you could also see that the connections to politics are getting stronger — from the startup scene to politics generally speaking, and in Berlin specifically to the Senate. I came home with a very positive feeling. A sense of new beginnings.
Jan Thomas: I think Berlin is great, this is by no means Berlin-bashing, but I don't like this competition between Berlin and Munich at all. And above all, I thought Kai Wegner — again, I don't want to start bashing — but I found him quite lackluster. He basically only articulated the demand that Berlin belongs on top, but hasn't said how he wants to make that happen. And I had the feeling he was engaging with the startup scene for the very first time. At the same time, he was there for a long time, you have to give him credit for that. He's been awakened at least, I think.
Björn Rieckhoff: Yes, I think the presence alone helps already, quite a lot is achieved with that. I agree with you insofar as I don't entirely share this small-state perspective either. Honestly, I couldn't care less whether it's Berlin, Munich, Hamburg, or Cologne. What matters to me is that Germany builds a strong ecosystem overall and, in a broader sense, pushes Europe forward. I say that of course as a patriotic European, as I've appeared in recent episodes. But I think the more you see such community initiatives — it's called Berlin auf die Eins and we do belong there in my opinion and it does have the ambition to be the flagship. But it's not deliberately saying that everyone else shouldn't matter. No, we all want to support each other. That became clear in the discussion on stage, I think.
Jan Thomas: I just remembered Andy Möller, the football player. "Milan or Madrid, as long as it's Italy" was his famous line. And I feel it's a bit like Berlin or Munich, as long as it's Europe. I think that's how I'd summarize it for myself. But I thought, maybe we should briefly mention the energy in the room and especially what Julian Teicke is building there. He was also a guest on my podcast, former WeFox founder, now he's building the Delta Campus. That's pretty cool, I must say.
Björn Rieckhoff: Yes, of course he also had WeFox in the past, so that backstory naturally provides a certain leverage. It's different from me sitting down and trying to initiate a community. That has a different altitude from the start and it's good for the scene. That actual scene leaders — not just self-proclaimed ones as so often — are taking the initiative and leading by example. So hats off, or caps off.
Jan Thomas: And also interesting, since we're talking about Munich, some people from the Young Founders Network were there, I keep running into them at the Delta Campus and I found it fascinating — they came up from Munich specifically for the event and drove back the same night. That's impressive, you have to say. It's a signal. And it also shows that perhaps Munich and Berlin are closer than you might think. Maybe they're simply two different locations that in symbiosis create something special and you shouldn't play them against each other.
Björn Rieckhoff: Absolutely. And I think being just one ICE Sprinter connection apart really isn't that far. And with that in mind, I think you can do a lot more together. And maybe Kai Wegner and Markus Söder should sit down at the same table. I would actually really like to see that.
Jan Thomas: I'd like to be there too, over a Weißwurst and a Karneval. But tell me, you also told me you're heading straight to another event and I think we need to expand on that a bit, because you're actually fully immersed in this whole OpenClaw world.
Björn Rieckhoff: Yes, OpenClaw or, in a broader context, how AI can help me build out my service business, and that really hits a nerve. So you just mentioned it, the hook was a current event — Point9 is hosting an OpenClaw meetup on Friday in Berlin. Christoph Janz had shared it multiple times on LinkedIn. I think they're at 280 RSVPs now, so quite large.
Jan Thomas: We can link that too. It's a public event.
Björn Rieckhoff: And they basically build all afternoon. There are mentors from OpenClaw, Postalk, OpenAI and others giving feedback and so on. Disclaimer, I'm not involved in organizing the event. I'm just jumping on because I thought the initiative was great. I actually haven't used OpenClaw that intensively myself yet. But that's partly because I've been doing well with a different setup. I'm building myself an Advisory OS in the background, because I can only get so far with the standard versions of OpenAI, Anthropic and others for what I do. And I just need a few research agents in the background that help me analyze things, but also good API integrations that help me keep lists up to date and enrich them. And that's much easier to build in an interface. I've actually been thinking about whether, once it's done, I might share some insights and publish a GitHub repository. The reason I'm diving into this — and I've done VC for the last ten years, I'm not a techie myself — but I think it's extremely important that we engage with this, from a position where I know that in the future I won't be able to earn money as easily. When you offer advisory services, the knowledge will at least get commoditized and then it's really about the judgment question, about context, experiential knowledge, and also relationships with founders and investors. And with that in mind, I'd rather cannibalize my own advisory business and develop something new from it than be overtaken by the market. So maybe a GitHub repo can help others operate more professionally too. That would be nice.
Jan Thomas: Totally. But I find that really interesting, I have to say. And of course, you're under pressure right now, as you're describing, because the market is going to change somehow. You can maybe only feel it in the early stages so far. But it can come with full force. And then it's interesting, independent of you, to think about the entire consulting industry — what will it look like in the future?
Björn Rieckhoff: Yes, exactly. I recently read a great blog post, we can link it too. Sequoia published on March 5th, Julian Beck wrote it — Services, the New Software. And he basically argues that hyper-specialized services will lead the territory in the future and trillion-dollar companies will emerge from it. Compared to a company that used to sell software, the shift is toward selling services holistically. And he created a great market map that divides individual industries into Intelligence and Judgment, and on one axis into Insourced and Outsourced. We can link that, it's super interesting. And as a mental model, I think it's very interesting for founders and investors to see.
Jan Thomas: I've already opened it. Didn't know the article, but sounds super interesting. Are you still worried overall though? I'm wondering, in the industry you're in, what's the core you can still compete with? And you just said Judgment and things like that. Is it then network, access, and trust? Are those the things you need to strengthen, and do you perhaps need to start building more of a personal brand in certain areas today?
Björn Rieckhoff: Absolutely. I mean, I'm not sitting here with you every month for nothing. I also enjoy it a lot. But it's of course important to have a certain visibility. And I think what helps is building up experiential knowledge yourself. And the models are getting better too. Critically examining things and above all checking when which proposal is appropriate and implementable. And I agree with you, much of it will be commoditized. What I'm thinking most about right now is that in consulting businesses, the logic has historically been billing by the hour. Similar to lawyers. And you'd sell the consultant day. But if I can now accomplish significantly more in the same time or conversely, if I need significantly less time for a certain analysis, then that's an absolute price collapse. So I'm currently strongly moving toward value-based pricing and really want to get away from hourly packages or retainers and the like. I think we'll see how that works. My approach is basically Build in Public. I talk very openly about challenges because, from what I've learned, that gets me ahead fastest.
Jan Thomas: Yes, but isn't this business model already a bit with its back against the wall anyway? As they say, once the toothpaste is out of the tube, you can't get it back in. So maybe that's what's happening right now with the AI wave. It's just real.
Björn Rieckhoff: It is. And it's becoming more and more critical infrastructure, to slowly transition.
Jan Thomas: I was just about to say, that's a great bridge, exactly.
Björn Rieckhoff: Exactly, finding the segue to the current topic we wanted to discuss. Let me just jump in. You mentioned this on your daily podcast too. Nscale announced a massive funding round, two billion dollars Series C. They say it's the largest Series C ever in Europe. The company is based in London. And what the company does is essentially provide AI compute on the basis of large GPU clusters. So you have this classic value chain: chip manufacturers, data centers, then cloud and platform providers, and after that the models from OpenAI, Anthropic and others that become applications. And they basically build this entire base of data center plus services on which providers can train their models and book capacity. And the closest comparison is a US company called CoreWeave. They're also publicly listed now. The stock hasn't performed that great in recent months, but I think it's a very good comparison. They went public at 23 billion dollars. For comparison, we now have a valuation here of 14 billion. Almost double. CoreWeave generates around 5 billion dollars in revenue according to Reuters. So already very, very large.
Björn Rieckhoff: What I find so fascinating about this topic is simply the significance of this funding round. Not just in terms of size, but that you're entering territory where you see that AI infrastructure is no longer this classic software business or a startup game, but is actually infrastructure. It's an infrastructure project that's also treated like one. Not just with classic equity financing, but with collateralized loans. And the collateral is based on the GPUs. And that's so cool, I think, when you read into it. You now have GPUs, compute power, physical data centers as assets that are simply bankable, because banks see this sustainable value in the business model or in the asset and lend against it. And not sparingly, but with very large debt packages. And that shifts the perspective from, we have a quick venture capital round here, to, we have a sustainable infrastructure project.
Jan Thomas: I've been reading up on it and trying to grasp it. I find it absolutely fascinating, I have to say. I always look at it and think, if someone had pitched this to me at the beginning, I would have said it'll never work. I don't know CoreWeave in detail, but I thought it sounds so complex what they have to manage. It's not just a simple cloud provider or something. There's so much attached to it. All the way to sales, all the way to — you just said it's bankable — and I'd say let's discuss in detail under what conditions it's bankable. But I find it so complex. Of course you eventually build moats because it's so complex. Do you know what I mean?
Björn Rieckhoff: Yes, it's interesting because I think the question is always a bit, what are the difficulties with this model. Obviously it's incredibly capital-intensive, you have to build physical data centers, and at the same time you need a lot of knowledge when it comes to things like not just orchestrating the hardware itself, but also topics like energy procurement.
Jan Thomas: Exactly, that's what I mean.
Björn Rieckhoff: And the company, as I understand it, actually emerged from the crypto mining space. And that was back when everyone basically woke up to GPUs.
Jan Thomas: That's how Nvidia got big too, you could say.
Björn Rieckhoff: Exactly, that GPUs aren't just relevant in gaming but also in modern compute operations for crypto mining and then AI — very close together — can operate much more efficiently than a CPU. And I agree with you, you very quickly fall into thinking, oh God, that's such a thick board to drill through. But I think especially in current times, where there's a very high willingness from investors to provide the necessary capital and where you can build this narrative very strongly that this is sovereign infrastructure within Europe, you can then move the kind of capital that also gives you the drill to go through that thick board.
Jan Thomas: But in a market that constantly has tailwind. So that's exactly why I'd put a question mark on this whole bankable thing. I don't want to talk an AI bubble into existence, but we're currently in a market that's really driven by greed, where everyone says, let's ride the front of the wave and build another data center, another data center. We don't even know yet if the market in two years will really behave as anticipated. And then you have massive capex here. You have giant data centers and long-term contracts. So the company is actually relatively inflexible afterwards in terms of responding to market conditions.
Björn Rieckhoff: Yes, that's right. I think the current view is, and I looked into this, I believe there was a Reuters report from early March: they expect the big tech players to spend a combined 630 billion dollars on AI infrastructure. Chips, servers, networking, data centers. 630 billion dollars. That's of course many times higher than individual funding rounds of two billion. And from that we can see the scale this topic can grow to. And at that scale, this is actually more like a small fish, almost, when you see it in relation. And in parallel, there are consulting studies that say AI-ready data centers are still too few and you need more capacity to keep up with the massive expansion of AI adoption.
Jan Thomas: Remember in quick commerce there were always these companies that started renting out riders or matchmaking drivers because they were a scarce resource. So I don't want to badmouth Nscale here, I think it's totally cool, including that they're in Europe and so — lots of checkmarks. But it's all going a tick too fast for me and it's always the largest Series C ever. A lot of money is flowing right now.
Björn Rieckhoff: I feel the same. It's in the range where you're amazed and at the same time a bit uneasy, because you get the impression that things are slightly getting out of hand or becoming hard to grasp. When I started working in VC in 2016, a 5-million funding round was a lot of money to me. And now we're talking about billion-dollar funding rounds. And I think it takes some time to orient yourself to these magnitudes. At the same time, you get the impression that developments are so dynamic that you can't keep up. Like, to put it simply, the discussion we had about which tech stack to use, do I go with OpenClaw for agents or do I build this on an AWS instance in Frankfurt. These are decisions you make and you're not even sure if you'll have to reverse them in a month because some model comes along that runs much more efficiently on different infrastructure. And that's the crux of it — you can't quite anticipate the path forward. And at the same time, you have this professionalization in the market where these are no longer tech bets but infrastructure projects, like Nscale.
Jan Thomas: Yes, and that there are many use cases for AI is completely clear. That they'll all eventually become reality — take video, that AI plays a role in video, that AI changes music, changes programming, Waymo and others will come — all totally clear. The question for me is a bit the timeline. And whether players like these — you're totally right, two billion is then peanuts compared to the 600 billion you just mentioned. But these are all magnitudes that come with such force and nobody knows how fast the adoption speed will be. There's a missing gut feeling for where the market is headed, because we've never had a time like this.
Björn Rieckhoff: Yes, that's absolutely right. I think you're making the bear case for the investment there. So there's certainly a capex and leverage risk that you're taking on. Obviously, something like this scales quickly through large debt packages. You can then build the hardware. And CoreWeave has shown this too, I think. I read a couple of articles — they've repeatedly used very strong debt financings for expansion. However, the model is of course super sensitive to demand dips. And then you very quickly have a situation where, within a certain refinancing window you need to hit, you have to show the corresponding growth to be able to retire the obligations. And I do have the impression that we're sailing very close to the wind. And in a way rightly so, but that statement is of course entirely based on a gut feeling. That there's this massive trend, but to what extent it's being overstretched or not is extremely hard to assess.
Jan Thomas: I mean, the fact that they're delivering — again, Europe, I totally celebrate that. That feels good, and that they've managed to put such a round together, also impressive. I think the investors, let's maybe briefly go through them — there was Astra, Dell and NVIDIA, and then Nokia and Lenovo. A lot of hardware, I feel, driving the whole thing. That's cool but also very strategic already. But then they've also assembled, I think, a really cool board. Sheryl Sandberg, I haven't read about her in a long time, she's joined for example. So something is happening there, and to have put together a round like this, they must have already delivered. The traction must be right.
Björn Rieckhoff: Absolutely, I would assume so. I think the financing history speaks for itself. They had, I believe September 2025, raised 1.1 billion. Then through strategic investors there was another kind of SAFE note, which was also 430 million dollars. But that was actually a SAFE note and after that came a 1.4 billion financing round that was GPU-secured. So these two billion being communicated now — it's not just equity financing, that's very important. But you can see in total how much capital is flowing in. As you mentioned, these aren't classic tech investors anymore, they're infrastructure investors. And I think the lead investor you mentioned from Norway, they actually come from the energy sector. And you can already see that it's about location factors. And recently Emmanuel Macron said — well, he has nothing to do with Norway, but we can all place him — he said for example that France's big location advantage over other countries, nodding toward Germany, is that they have all the nuclear power plants and can actually provide the necessary power when they scale everything up, compared to other nations. And that's of course very quickly at a level where it becomes political, where it's a location factor and a strategic decision for the country.
Jan Thomas: I've now opened both cap tables on Crunchbase. Nscale has actually already raised 5 billion in total, that's quite something. But you don't see a single typical growth investor that we know from the traditional startup scene. No Andreessen Horowitz or Sequoia or anything like that. But that wasn't the case with CoreWeave either, which is also interesting. The only well-known startup VC there is Coatue. And then Blackstone and Fidelity and such, Goldman Sachs. So maybe they're also generally optimized for different types of capital raising, that could be it.
Björn Rieckhoff: Yes, that could well be. And at the same time, the way you finance this is naturally not the classic VC approach anymore. When it comes to bankable securities, you're often in a different arena and I have to say, from my perspective, actually a more sustainable one, because you have physical assets behind it. At the end of the day though, it's also a business model where classic tech investors — I don't want to say they've lost too much — but where there are many topics they simply have no expertise in. When it comes to power and cooling technology or building networks and cloud services and the like, that goes way beyond software. Can of course be wonderfully supported by software. But you need real-world know-how somewhere. And there are physical constraints in the build-out. And I think many VCs are simply more cautious when it comes to such models. But that doesn't mean they can't scale accordingly.
Jan Thomas: Yes, I'm also saying that because I was looking for the catch, because of these somewhat unfamiliar investors. But maybe that's just how it is in this space. What just occurred to me again — you might have mentioned it earlier — but Nscale was only founded in May 2024. That's crazy. So I'm just wondering, do you think this is a market that's still emerging? Could new players even enter now?
Björn Rieckhoff: Well, if I think about the spend from large tech companies we mentioned earlier — 630 billion dollars in investment in these areas — and of course you have the Metas of this world building their own data centers, but you also have the Apples of this world that are completely holding back. And I actually find that interesting to consider, because I think the bet is simply that they're saying, why should we put such assets on our balance sheet that need to be heavily depreciated over time. And I think we've discussed this before, that the big corporations massively overestimate the real useful life to present the depreciation favorably for their own balance sheet. And there are still large companies like Apple that don't participate in that game at all, or only to a necessary extent. And they'll probably approach players like Nscale and say, now we're ready and we have someone here we can reliably advance our strategy with. I believe there's room for more companies in this space. Especially because, as I understand it, Nscale is currently the only one in Europe trying to build this at scale.
Jan Thomas: What do you think the exit channel will be? Setting aside an IPO. Could you even imagine CoreWeave buying them?
Björn Rieckhoff: I don't think so. That doesn't fit the story they want to build with European infrastructure.
Jan Thomas: I was thinking precisely because of that — you know, as a US player you'd probably have a hard time achieving the same credibility in Europe.
Björn Rieckhoff: That's true. But honestly, I see more of a — I don't want to say a moat that can't be crossed — but I think this permeability that existed in the past, as a natural exit channel to the US because it opened market access to Europe for Americans and to America for Europeans, I think that's no longer a given in the future. Especially with topics that trigger a certain political interest, I think it will be more about bringing the Nscales of this world public in Europe and then financing them further accordingly. I don't really believe in the story of early consolidation here. Rather, the market will likely keep expanding. And there'll be more players. The market is simply too large to start consolidating already.
Jan Thomas: Yes, it would be great if there's eventually a really good IPO with a company that could be worth 50 to 100 billion in perspective.
Björn Rieckhoff: I've read that there are already indications of an IPO direction. They've apparently already mandated Goldman and JP Morgan as underwriters, but without a known timeline as far as I've read. But as for exit channel, I think the IPO is the only possibility at this scale at some point. And side note, even if you've mandated such large banks as underwriters, that doesn't mean there won't be another funding round or two in the private markets in the meantime. As you showed earlier, given the large volumes flowing in the US, it's quite a valid scenario to simply stay private for a while.
Jan Thomas: Yes, it's going too fast for me. I'd actually find staying private more appealing right now, because we all hope that the IPO window opens again and that we see more IPOs. But they have to be the right ones. And if now with this greed I mentioned earlier, too many companies that aren't really IPO-ready go public too early, then you'll experience another shock. And that would be bad for the market, I think.
Björn Rieckhoff: I'm absolutely with you there, if that's how it turns out.
Jan Thomas: It's not a bad company because of that, but if you could wait another 2-3 years and take a solid company public, that's better in many cases I think.
Björn Rieckhoff: I'd say more conservative investors would probably like to see more than one full fiscal year in the books, which was 2025. You mentioned earlier, it was only founded as a spin-out in 2024, so you have a stub year and then one full year and then it's supposed to go public at some point. That's maybe a bit thin to build an IPO story on. So I could well imagine that 2026 will pass and then sometime in 2027 plans get more concrete. But I think what's happening is that the shareholders and management probably don't want to miss an opening IPO window, so they're positioning broadly enough early enough to simply have optionality.
Jan Thomas: Cool, Björn. I thought that was great. And as I said, bottom line — first of all it's great that such a company now exists in Europe too. That's exactly what we wish for. More large companies that are shaping this AI wave, with a European counterpart in each case. I thought it was a shame that Mistral seems to have faded a bit. We've discussed Aleph Alpha here before too. You just want more champions here right now.
Björn Rieckhoff: Absolutely. I mean, if I were to look on the LLM side and someone gave me an API that's just as capable as those from Anthropic or OpenAI, I'd gladly take it from Europe in a heartbeat. But unfortunately, the performance isn't quite comparable. And with that in mind — this closes the circle to our earlier discussion — I'm absolutely with you, if there were more such companies. I think there's also a great willingness from European companies and market participants to prefer the corresponding European solutions.
Jan Thomas: Maybe today was the beginning of a trend. Let's hope so. Björn, it was fun. Thanks so much and see you soon!
Björn Rieckhoff: Until then, bye.
About Björn Rieckhoff
Björn Rieckhoff is an independent advisor and business angel with nearly ten years of experience in early-stage venture capital. He helped build Cavalry Ventures as its first employee and later became a partner of the fund. Today he supports founders more directly with fundraising — sharpening their story, stress-testing business models, and setting up lean financing processes. With over 80 transactions and board seats from seed to Series B, he brings this perspective as a sparring partner for entrepreneurs.
About Startup Insider
Startup Insider is the industry portal for the startup scene in the DACH region. It covers news from all regions and industries, along with an overview of key players and events in the German-speaking startup world.