Netcompany Group A/S (CPH:NETC)
Denmark flag Denmark · Delayed Price · Currency is DKK
364.60
+0.40 (0.11%)
Nov 16, 2026, 5:00 PM CET
← View all transcripts

Investor update

Nov 10, 2026

Summary

A DKK 325m investment secures a 23.7% stake in DCAI and positions the partnership to deliver a sovereign AI stack from compute to applications. Denmark’s projected 2027–2031 sovereign AI opportunity exceeds DKK 30bn, with revenue expected from 2027.

Operator

Welcome to Netcompany's presentation on delivering and scaling the future of sovereign AI in Europe. Today's call is being recorded. If you have any objections, please disconnect your line. All participants will be in a listen-only mode throughout the presentation. Afterwards, there will be a question and answer session. Today's speakers are CEO André Rogaczewski and CFO Thomas Johansen. André, please begin.

André Rogaczewski
CEO, Netcompany Group

Good day and welcome. My name is André Rogaczewski, and I am the CEO and Co-Founder of Netcompany. I am joined today by our CFO, Thomas Johansen. The investment we are announcing today is about more than just an investment in computing power. It is about whether Denmark and Europe can develop and deploy AI on our own terms, including in the parts of society where security and trust matter the most. With Gefion, we already have the foundation. Now we want to take the next step. We want to bring that technology together with the platforms, integrations, and practical solutions needed to create the real value from AI. I will start with the investment itself, explain why this is particularly interesting, and what partnership that makes a difference here, then show the opportunities it creates for society and for Netcompany.

Before we get going, there are some important disclosures that I need you to read through, so could we please have slide number three? I will pause here for 30 seconds and let you have a read through all these important disclosures. With that, can we please go to slide number four? Netcompany is investing DKK 325 million in DCAI through newly issued shares, giving us a 23.7% ownership stake. DCAI operates Gefion, the Danish foundation for sovereign AI compute. Our stake will be the same as EIFO's, and we will have the right to appoint one of the five members of the Board in DCAI. What makes the investment exceptional is its partners, the Novo Nordisk Foundation, EIFO representing the Danish state, and Netcompany. Together, we will bring long-term commitment, sovereign infrastructure, and the capability to deliver mission-critical services. This is only the beginning.

We expect AI demand to grow exponentially over the coming years, so DCAI will need substantial additional funding to expand compute capacity and operations. DCAI is to secure that growth capital through third-party financing as customer demand develops, and it will not rely on repeated capital contributions from Netcompany. Can we please go to slide number five? Technology is increasingly a question of security, competitiveness, and control. Europe cannot be content merely to use AI. We must also be able to determine how our most critical solutions are operated, governed, and developed. Sovereign AI is not simply a supercomputer or a national model. Compute, models, governance, integration, and applications must work as one system. Otherwise, we end up with promising but very isolated initiatives.

We need shared platforms that can be reused across use cases, individual authorities, and suppliers should not have to rebuild the infrastructure, security, and model controls each time they put an agent into production. The Novo Nordisk Foundation and EIFO established the foundation with Gefion. Netcompany brings experience in the building and operating mission-critical IT, secure integration into government systems, and Feniks AI to turn computing power into usable services. Our ambition is to be the first to prove that a complete sovereign AI stack can support critical services at high scale, and then take that experience and those solutions across Europe. Can we please go to slide number six? What does it actually take to establish a national AI capability? Our answer is five pillars that must work together, not five independent projects. At the bottom, we find the infrastructure.

That's the data centers, the GPUs, networks, and compute. Gefion and DCAI provide that foundation. Above it sits the models. Our approach is open and model-agnostic, so we are not dependent on one single model provider. The third pillar is the AI platform. Feniks Governance brings model orchestration, security, access control, metering, and governance into one shared services at this layer. Our ambition is for Feniks Governance to become the default platform, so every authority and vendor does not have to rebuild the whole stack from scratch. The fourth pillar is government integration, connecting agents securely to registers, existing applications, and trusted public sector data. Netcompany is here uniquely positioned because we already build and operate a very substantial share of Denmark's mission-critical central and local government systems. We know what it takes to integrate them responsibly.

At the top, we find the agents and applications that deliver real outcomes. Taken together, the five pillars form one operational sovereign AI stack. Our conviction is that no other company in Denmark, or in Europe for that matter, can combine this depth of government integration, trusted operations, and deployable AI services with the same reach, especially now with the DCAI partnership behind it. Can we please go to slide number seven? So Feniks is not a single AI product. It is a software suite designed to move customers from model access to working services on one secure, reusable foundation. The aim is to avoid an integration and governance project for every new use case. Feniks Governance is the common control plane, model selection, inference, user access, security, governance, monitoring, and usage-based billing.

It gives authorities and other suppliers the shared framework they need to adopt AI easily and responsibly. Feniks Learn helps us to understand existing, often complex landscapes, mapping legacy applications, business logic, and dependencies, and supporting the analysis so you get very fast from planning to acting. Feniks Build supports AI-assisted software engineering and the implementation of new solutions. Its purpose is to accelerate development while keeping it within the quality and security standards required for critical systems. Feniks Test helps automate test design, execution, and regression testing. The objective is to shorten release cycles without compromising the reliability that government systems require. Together, Feniks Learn, Build, and Test enable us to fundamentally transform the software development cycle, from understanding complex legacy systems to developing, testing, and deploying new solutions on a trusted and secure platform. Feniks Agentic takes us from AI infrastructure to AI that scales in practice.

It enables us to develop and deploy ready-to-use agents that solve real operational challenges, while Feniks Public Sector provides the domain knowledge and understanding of Danish legislation and public administration needed to make these solutions relevant and reliable in the public sector. Together, they work as a catalyst. Imagine an agent developed to solve a real administrative problem for one Danish municipality or authority. Rather than stopping at a proof of concept, that capability can be reused and adapted by other municipalities or authorities, moving from one local pilot to national scale deployment far faster than traditionally has been possible. Crucially, this is an open platform. We want other IT suppliers, not only Netcompany, to build and deliver solutions through the same sovereign foundations. If the full supply ecosystem can reuse common capabilities, the whole of Denmark can scale AI rather than everyone creating their own isolated stacks.

Proving this model in Denmark, we will have real practical experience, references, and solutions to take it on to other European markets facing the very same challenges. Can we, please go to slide number eight. The real opportunity is not simply to make AI available. It is to make AI dependable for the functions that governments, citizens, and regulated business rely on. Those are environments where control over data, technology, and operations is part of the service itself. That is why the distinction between different levels of sovereignty matters. The CSF framework sets out five levels of cloud sovereignty. A large part of the general AI market will be served at the lower CSF levels. Our ambition is to serve the critical public sector and regulated sectors workloads that demand higher levels, particularly CSL 3 and CSL level 4.

This is exactly what differentiates our proposition from the general AI market. That ambition builds on capabilities Netcompany already has, because we understand the registers, the legacy systems, security requirements, and the day-to-day operations behind many of Denmark's most important public services. Combined with Gefion and Feniks AI, that experience enables us to turn sovereign compute into solutions that work within existing government systems. Because we are model agnostic, we can choose fit-for-purpose, open-source, and open-weight models rather than becoming dependent on any single provider. It is also a more sustainable way to deploy AI. Shared infrastructure and appropriately sized models can help us avoid unnecessary training, capital expenditure, and energy consumption per task, while preserving the quality and reliability that critical systems demand. The combination of sovereignty, practical integration, and efficient operations is where we believe Netcompany will make a real difference.

Netcompany is uniquely positioned by combining sovereign infrastructure, European control and our proven ability to deliver mission-critical IT, we believe we can turn this ambition into reality and establish a model that scales from Denmark across Europe. I would now like to hand over to our CFO, Thomas Johansen, who will take us through the economics of the investment and the size of the market we believe it can unlock. Can we go, please, to slide number nine? Please go ahead, Thomas.

Thomas Johansen
CFO, Netcompany Group

Thank you, André. Let me put some numbers on this very exciting strategic opportunity. The investment gives Netcompany exposure to sovereign compute, but it also positions us to participate in the orchestration, government integration, and AI applications that will turn the infrastructure into services people and organizations actually use. Importantly, we are building on capabilities and customer relationships that Netcompany already have today. We see four economic streams in the transaction. The first is DCAI's standalone compute business, where Netcompany participates through our equity stake. The second is orchestration and governance through the Feniks Governance platform. The third is business services, including our products, platforms, and secure integration with government systems and data. The fourth is Feniks applications and agents that solve concrete operational challenges. The last three are direct commercial opportunities for Netcompany, in addition to the DCAI compute business.

We already deliver and operate a substantial part of Denmark's mission-critical public sector IT. We understand how to integrate with existing registers, application, and workflows. Through Feniks AI, we can make those capabilities reusable for authorities and other suppliers. That combination of trusted execution, integration, and common AI platform is what we believe can make us uniquely competitive, and it is a value proposition that will resonate in all of our markets. We are deliberately conservative in how we size the opportunity. By 2031, we estimate 50 MW - 80 MW of sovereign AI IT load in Denmark, compared to 200 MW - 320 MW for the total Danish AI market. We are comfortable with the direction and order of magnitude of those planning estimates, and as said, we view these as conservative. Our revenue assumption is also conservative at around DKK 100 million per megawatt of IT load per year.

Capacity will be deployed and utilized over time. Across the period from 2027 through to 2031, our model indicates more than DKK 15 billion of compute market revenue. We believe that the service layer above compute can represent at least another DKK 15 billion, and most likely more. Together, that is a market opportunity exceeding DKK 30 billion over a five-year period. These estimates are for the sovereign market in Denmark alone. Our investment in DCAI puts us in a strong position to compete across the full value chain, from infrastructure and governance to applications that create value in the real economy, and hence tap into that DKK 30 billion opportunity. The reason we are making this investment is much bigger than owning a part of the supercomputer.

Together with our partners, we want to make sovereign AI work in practice for Denmark first, and then at European scale, which will be accretive to our growth and margin ambitions. With that, we will conclude our presentation this morning and be happy to take questions. Let's go to slide number 10 and open the call for Q and A.

Operator

Thank you. If you do wish to ask a question, please press five star on your telephone keypad. To withdraw your question, you may do so by pressing five star again. We will have a brief pause while questions are being registered. The first question is from the line of Yiwei Zhou from SEB. Please go ahead. Your line will now be unmuted.

Yiwei Zhou
Analyst, SEB

Hi, André and Thomas. Thank you for taking my question, and congrats to the transaction. Two questions from my side. Firstly, can you talk about if there's any limitations to use Gefion to European Commission or other European countries? If there is, what are the limitations? Secondly, could you maybe elaborate a bit more about your expected timing of realization of the revenue stream for the coming years?

André Rogaczewski
CEO, Netcompany Group

Thank you for those two questions. The short answer to your first question is there's basically no limitation. I mean, we have a software suite that we put on the infrastructure of DCAI, and we can serve all regulated entities in Denmark, but also in Europe. When it comes to realizing some of this, we will start immediately. We are deploying all our software. We're going to our customers. We'll start seeing the effects next year, from 2027.

Yiwei Zhou
Analyst, SEB

I think

André Rogaczewski
CEO, Netcompany Group

Good question.

Yiwei Zhou
Analyst, SEB

My second question.

Thomas Johansen
CFO, Netcompany Group

Second question, André answered the way, and that is that we will start to see revenue from this already from 2027, and then it will accelerate.

Yiwei Zhou
Analyst, SEB

Okay, thanks. I will jump back to the queue.

Operator

The next question is from the line of Aditya Buddhavarapu from Bank of America. Please go ahead. Your line will now be unmuted.

Aditya Buddhavarapu
Analyst, Bank of America

Hey, good morning, André, Thomas. Thanks for taking my questions. Just a few from my side. Firstly, could you talk about for DCAI, what are the customer commitments that they already have in terms of contracted capacity? Could you talk about the pipeline they have as well? Then maybe just how the business is doing currently. Second, when you talk about those four revenue streams, how should we think about how those revenue streams, the economics of those, would there be some sort of revenue share between DCAI and Netcompany? Maybe talk about how that would work in terms of the relationship with those end customers. Finally, you spoke about this being model agnostic. Does this change the economics of the token cost, et cetera, for Netcompany and maybe your own internal training efforts or AI capability efforts as well?

André Rogaczewski
CEO, Netcompany Group

Thank you for those questions. Really interesting. I'll take them one at a time. When it comes to DCAI and the pipeline and how DCAI is performing, if you saw the press conference this morning, DCAI was started two years ago, but the last six months, it has really become a default choice for many that want to source GPUs and infrastructure. The raw metal sale, you can say, is really picking up at DCAI, and the pipeline is looking really promising. When it comes to the revenue streams, we have four, as I mentioned before. The orchestration and governance part will come default as the bottom layer of everything we do, and that we will, of course, sell almost to anyone who wants our services. That's almost a prerequisite.

Then on top of that, we'll put all the Feniks things you saw, all the services and the Feniks products, except for Feniks Agent. You will see them operating on that revenue stream. So every time we sell a Feniks Learn or we sell Feniks Build to develop software, or we sell Feniks Test in order to make more rapid tests, we will be using both the governance layer, but also, of course, the raw metal. Finally, maybe most interesting, Feniks Agent will be deployed as a real software layer where you can define agents working inside regulated systems that we will expose using MCP-enabled interfaces. So for anyone working in a regulated industry, you will be able to set up agents working with your data in a secure environment, just like you do yourself when you're using an Anthropic agent.

Now finally, everything we do will run on this model-agnostic partition, and that means we can shift out the models anytime we want. We can find the models that are the most price-friendly in terms of what type of tasks you want to have accomplished with your AI. From a token price perspective, there is no doubt that this will become a real differentiator compared to the global frontier models that are increasing in price at the moment. I hope that answered most of your questions. Thomas, do you want to add something to this?

Thomas Johansen
CFO, Netcompany Group

Yeah, just to add to your second question, Aditya, and thanks for that also, in terms of the revenue types and lines and what have we. I think it is important to understand that there will be multiple revenue sources for Netcompany. There will be the one that we call the compute through DCAI, which is where we participate with our equity share, but we will also have revenue compared from that model based on the software that will be utilized to generate the value in DCAI. That is one level. Then there is all the models and all the products and platforms that André was discussing just previously, which will generate revenue to Netcompany also, both in terms of services, but also in terms of licenses. There is going to be a multitude of different revenue streams coming from this transaction.

Aditya Buddhavarapu
Analyst, Bank of America

Thanks so much. Just to clarify on that last point, when your software is used to generate value for DCAI, does that mean they pay you some fees? Actually, when your own services, like Feniks Learn, when those are used for external customers, is there a revenue share with DCAI or any of those?

Thomas Johansen
CFO, Netcompany Group

Without being able to disclose the details of how we have set it up with DCAI, then I will say this much, that there are multiple ways of revenue sources for Netcompany and also multiple ways of value generation in DCAI. Basically, both parties win from this.

Aditya Buddhavarapu
Analyst, Bank of America

All right, understood. Thank you.

Operator

Next up, we have Claus Almer from Nordea. Please go ahead. Your line will now be unmuted.

Claus Almer
Analyst, Nordea

Thank you. Yeah, also some questions from my side. First of all, congratulations with this interesting opportunity for Netcompany. Maybe we can start with why were you able to enter the ownership of Gefion? It looks like a quite attractive opportunity. Is it due to Netcompany's strong stand in the public sector, being a front-leading IT consulting firm, or? Yeah, that would be the first one.

André Rogaczewski
CEO, Netcompany Group

I mean, also if you listen to the press conference this morning, there's a strong urge to scale the usage of AI in order to get all the benefits in society. In that sense, we play an important role since we are already servicing more than half of the public sector with some of the more critical systems. Furthermore, we have the tools and the software, both across the software development cycle, but also when it comes to providing governments and institutions and other regulated businesses with agentic software and our possibilities. I think it's a combination of what you just said there, Claus.

Claus Almer
Analyst, Nordea

But, and then, that you could have offered anyway, whether you're an owner or not. So, it seems more like they wanted you to be a driver for the increased use of Gefion, as an example.

André Rogaczewski
CEO, Netcompany Group

I think when we said this morning that we'll be delivering the full AI stack, I really think you should look into that, because it is the full AI stack that is interesting for customers. You want the sovereignty, and you want all the capabilities up through the layers. And having control of that and knowing that and having a close partnership to that, and also being able to guarantee all levels of that stack is crucial when you want to go out and scale this. That's why. So, both parties are benefiting from this.

Claus Almer
Analyst, Nordea

Sure. So, I'm trying to hint that we really want to have these on Danish ground. Is it also important for the policymakers that it's also a Danish tech company that is facilitating all of these things for the public sector?

André Rogaczewski
CEO, Netcompany Group

Yes. That is absolutely important also from a competency standpoint and a sovereignty standpoint.

Claus Almer
Analyst, Nordea

Okay. Sounds good. My second question goes to the first projects we are going to see on Gefion with Netcompany participating. Would it mostly be existing projects that will be transferred to Gefion, maybe with a Feniks overlay, or would it more be totally new projects that is going to be developed?

André Rogaczewski
CEO, Netcompany Group

It is definitely going to be both. There is no reason why we should not be able to use agentic AI inside many of the regulated systems that we are already serving. Hopefully, and determinately, that will change and shape the way and the time spent on using these systems, and making them much more efficient in terms of serving the Danish population. Furthermore, when bidding for new systems and projects, this is also going to be a differentiator, absolutely.

Claus Almer
Analyst, Nordea

Okay. My final question, which I guess is for you, Thomas. You have a 2029 target, which is an organic target, as I recall. With the slide you had with this huge opportunity, should we add this to the midterm potential?

Thomas Johansen
CFO, Netcompany Group

We are already in 2029, Claus, getting ahead of us here. The way I can answer that is that this transaction will be accretive to our growth and margin ambitions. I am not going to discuss any long-term targets or anything else other than what we have already said. Clearly, this is a new and significant revenue pool we are tapping into, which did not exist when we discussed our targets a year and a half ago. That is why we are excited about it.

Claus Almer
Analyst, Nordea

Thank you so much. That was all for me.

Operator

Next up, we have a follow-up from Yiwei Zhou from SEB. Please go ahead. Your line will now be unmuted.

Yiwei Zhou
Analyst, SEB

Hi. It's Yiwei from SEB again. Two questions. Firstly, from the accounting perspective, Thomas, can you indicate how would you book the revenue streams, the full revenue streams? What would be booked above your EBIT line, and what would be booked in associated income? I'll do the next question later.

Thomas Johansen
CFO, Netcompany Group

Yeah. This is straight normal accounting. This is an associated company with which we do not have controlling majority. Therefore, the result of DCAI as such, as the associated company, will be below the line. Everything else will be above the line.

Yiwei Zhou
Analyst, SEB

Okay. For example, if you get a license income, let's say DKK 10 million, is it booked proportionally in the associated income, or would it be your sort of group revenue and EBIT?

Thomas Johansen
CFO, Netcompany Group

Group revenue and EBIT.

Yiwei Zhou
Analyst, SEB

Okay. Clear. Secondly, also, I'm trying to understand profitability of the data hosting and also associated AI computing service. I'm asking because the conventional data hosting business carry very low margin. I'm just trying to understand the margins of the AI data center. Could you give some indications here?

Thomas Johansen
CFO, Netcompany Group

We cannot discuss the expected margins in DCAI, but I think you make a very important distinction yourself, Yiwei, and that is to distinguish between normal data centers hosting and then what's going on in AI, as we are investing in here. It's two distinct different businesses. What is important here is that this is not so much just about getting a return on, quote, unquote, "the iron," i.e., the running of the GPUs. It's also wrapping in, and more importantly, and that is important to understand because this is really where all the value is being unlocked, it's wrapping in the services, the applications, the agents in a secure, sovereign manner, which today does not exist, neither in Denmark or in Europe. So if you try to compare, it doesn't exist.

You can see various competitors that might have two or three of the elements, but not all of them. So clearly what we are servicing here is something that has more than just, quote unquote, "just compute." Also, as André was alluding to, remember, we are only focusing on the high security level layers. So we are not competing with the hyperscalers in terms of who can make the most on SOMI, which is where a lot of the AI has been going on right now. This is really what is going to make real value for governments and for highly regulated businesses. Maybe you want to add there, André.

André Rogaczewski
CEO, Netcompany Group

Yeah, I think that what is important to understand here is that compared to traditional, I guess, hosting and software, which is very, very standardized, here you have the four different layers. It is also the mix in between the layers and the optimization in between the layers. So having control of the lower layer can make you change your LLM and then process much more efficiently and still take a good price. But also having control of your agentic capabilities can actually bring you to a state where you are much more efficient with your agents running on a cheaper LLM. So you need to control all four layers in order to get the maximum value out of hosting this, and that is what it makes it different from traditional hosting, and that is why this opportunity is extraordinary interesting.

Yiwei Zhou
Analyst, SEB

Great. That was very clear. If I may ask one more question, just one, could you confirm that if you were to develop or run a project system for defense sector customer, is it a prerequisite that it has to be run on the sovereign data centers?

André Rogaczewski
CEO, Netcompany Group

Yes. That is a prerequisite for most countries and also in Denmark. Yes.

Yiwei Zhou
Analyst, SEB

Thank you. I will jump back to the queue.

Operator

Let me just remind you, if you wish to ask a question, please press five star on your telephone keypad. As there are no further questions, I will hand it back to the speakers.

André Rogaczewski
CEO, Netcompany Group

Thank you all for joining in on this extraordinary call. Have a wonderful day.