Authors: Elena Vrabie and Bogdan Iordache
If you want to cut through procurement bureaucracy and get honest, practical insights about defence technology, go straight to the battlefield and test your theory, because “there is no fancy technology in the trenches when soldiers are under attack,” says Marc Wietfeld, co-founder and CEO of ARX Robotics, a German defence tech startup specializing in autonomous unmanned ground vehicles.
A mechanical engineer who spent over a decade as a German Army officer and participated in military robotics research, Marc co-founded ARX in 2022 alongside Stefan Roebel (COO) and Henrich Maximilian Wied (CFO).
On a mission to protect democracy and help NATO soldiers transition from shooters to system orchestrators, ARX Robotics has raised over €40M to develop its modular hardware-software platform. The company’s Gereon robots and Mithra OS transform both autonomous systems and legacy military vehicles into interconnected tools for reconnaissance, logistics, and crisis response.
In the interview below for Underline Ventures, Marc Wietfeld shares how battlefield experience drives practical defense tech design, emphasizing Europe’s need for technology sovereignty and the rapid scaling of autonomous systems to increase military effectiveness, rather than slowly and steadily later.
Underline Ventures: How does your personal experience, both on the battlefield and in tech research, shape the design and priorities of ARX’s products? Does this dual perspective lead to practical, soldier-focused innovations?
Marc Wietfeld: When we started ARX, a lot of investors had concerns because the founding team didn’t look like the typical robotics tech founding team. We are not MIT PhD graduates; we never went to Stanford. However, we understand the urgency and the pain, and we know that you cannot solve a problem you have never had. The heart of the project was to bring technology faster into the hands of the soldier.
Additionally, the infrastructure of armed forces is unique, and so is the market, so we aimed for easy execution. What we learned as officers and from the battlefield is that only the easy things, the ones that can help you, will function when shit hits the fan. There is no fancy technology in the trenches when soldiers are under attack.
As an infantry soldier, your preparation is tailor-made to the mission. Is it reconnaissance, combat, or logistics? Then you pack your backpack for, let’s say, a recon mission, and after two hours, it becomes a combat mission for three days, and you find yourself with completely the wrong things with you.
It was clear to us that everything we are doing should be easy, rugged, adaptable, and modular, because things are changing super fast in tactics and on the battlefield. In addition to that, what we also saw is that technology is changing even faster.
You’ve previously spoken about “credible deterrence” and rapid rearmament. From your perspective, what concrete measures should European governments take to build resilience and autonomy in defence tech, without escalating tensions?
MW: Europe has a lot of disadvantages when it comes to defence and resilience. We are outnumbered by our potential adversaries in both materiel and personnel. We are outpaced when it comes to the implementation of new technologies in the armed forces. To build an honest deterrence, you need technology sovereignty.
Geographically, we have a disadvantage when you attack Europe from the East, and we need to balance it. The only way to become stronger is to multiply our current force of one soldier to five soldiers as fast as possible through technology, autonomous systems, and affordable systems. This is why unmanned systems and autonomous software capabilities are needed to build deterrence.
We need new defence tech primes and startups to completely change the procurement cycle, because now it is made for traditional combat vehicles, like tanks, airplanes, or aircraft carriers. It’s fine to have a new fighter jet or tank generation every 10 or 20 years, but when it comes to new technologies, we need to be talking about weeks and months. So, a new procurement process to arm our military forces with new technologies, built by agile market players, is a must.
How ready and capable is ARX to deliver rapid, flexible robotic solutions that match the urgency of Europe’s new defence build-up? Can they realistically meet demand within a short timeframe?
MW: We learned a lot when the Ukraine conflict started. As a startup in the pre-seed stage, we donated 60% to 70% of our total capital to bring our robots as fast as possible into the hands of the Ukrainian Armed Forces, help them succeed, and learn how these systems gain a battlefield advantage.
We are prepared when it comes to the product, but it’s hard to scale in Europe as a defence company, because the procurement of drones is in small numbers, and drones are a numbers game. When Europe has to defend itself, we need drones not in the 1000s, but 100,000s, and the companies that can provide these numbers are not here due to the fact that you cannot scale alone, without the demand.
So, you need to build a company that can ramp up its production immediately, and for that, you need to answer a lot of hard questions. We choose a fully European supply chain and ownership because sovereignty is built on independence. So, a hard question is, can you build European sovereignty with Middle East money, because money means ownership? How can a company be reliant on overseas components, funding, and ownership to think and act European?
How does the nature of your product and its framework impact scalability, particularly with hardware constraints versus software flexibility?
MW: The core of scalability lies in the nature and the framework of your product. This is also why we struggle to build complex systems like tanks or airplanes, because there are a lot of special components involved.
ARX is a software and hardware company, and also a manufacturing one. The software part is highly scalable, but when it comes to hardware, you need a product or a framework of products that is hardware agnostic.
For example, if you only have one type of motor to drive your robots or only one type of battery your robot can work with, you will end up with a hard time scaling. Our software is capable of integrating components off the shelf. We can choose between over 100 types of motors, gears, or batteries to build our products.
Do ARX’s autonomous systems currently support battlefield operations like logistics, reconnaissance, and casualty evacuation? How do you envision the role of human operators evolving as AI and robotics technologies advance in these areas?
MW: Our main use cases in warfare right now are logistics and casualty evacuation of wounded soldiers in Ukraine. In Europe, it’s mainly reconnaissance and electronic warfare. We are working on fully autonomous robotic systems, but it’s clear that in the next few years, a human in the loop is required when it comes to battlefield decisions. I’m not talking about driving from A to B, even if it’s the battlefield or off-road, harsh terrain, but when it comes to decisions about life and death.
We are aware that our system accommodates and exploits AI. When AI is doing a step, our robots are doing five to 10 steps. We anticipate what AI will bring us in the next two to four years, and we are building the foundation to exploit this fast and to multiply ourselves.
We are at the point where one soldier can orchestrate five to 10 of our robots. He doesn’t have to waste his time and look at every video feed, or at a robot that drives 20 kilometers in the field. He can allocate his attention to the points where robots detect things and where battlefield decisions are needed.
How do you see the future role of modular robotic forces shaping deterrence strategies, especially in light of shifting force compositions and the importance of industrial capacity in modern warfare?
MW: I can refer to the Strategic Defence Review 2025 of the UK armed forces. It was published at the beginning of June, and I’m impressed. The UK understood that we would not have more tanks and more soldiers, but we needed a way to double and triple our lethality. We are active in the UK because there’s a shared mindset and strategy clarity.
What they say is 90% of our current lethality – which is something we would not dare to say in Germany, we would call it efficiency – but they call it what it is, is generated from survivable platforms, like tanks, helicopters, airplanes, aircraft carriers, the traditional platforms, and they will cut it to 20%. Having a split of its armed forces means 20% are the traditional platforms we know, 40% are reusable platforms, like our robotic systems, and the remaining 40% are systems not considered to come back.
Effective deterrence is made when you can compete with your enemies, and nobody in the world should be able to think that an attack on Europe could be successful. If you bring your few survivable platforms into a fight against 1000s of economic drones, you will not succeed on the battlefield, and you will also fail in the economy of warfare.
The European Armed Forces are great. They are well-trained and decisive. They will not be crushed in tactics or outmaneuvered on the battlefield. But if you look at history, the industry backbone often decides who the war winner is. We need to rebuild and feed the warfare of European armed forces when it’s needed.
What are your guiding principles for decision-making autonomy, especially as Mithra OS advances? How flexible and adaptable is Mithra OS when working with older or non-standard military equipment from different countries?
MW: Before we talk about autonomy, we should talk about digitization and modernization of the armed forces, because we are the first company to bring unmanned ground systems in larger numbers to the field. What we saw is that the new assets of warfare, like drones and software-defined systems, are not linked. They are not interoperable with the current armed forces. We are talking about 50,000 vehicles in the European NATO armed forces, which are analogous. Cars are most common from the 70s, 80s, and 90s. They cannot accommodate AI.
Currently, less than 5% of European armed forces comprise unmanned systems. But this will grow in the next decades to up to 50%, and it is only possible if you can link these two worlds together; otherwise, you will end up with a split armed force: new software-defined defence capable systems, and the old legacy fleets.
So, we found a way to turn these cars into software-defined systems and link them to one force. By doing so, you open up a kind of app store on these vehicles. If you have the vehicle software defined and capable of accommodating software and AI sensors, you can bring capabilities to it.
The guiding principle is that, as with the robots, we built the foundation to have the systems learn and exploit AI and machine learning to feed the newest capabilities to them when they are available. We are making them adaptable, updatable, and connectable to the new digitized warfare.
By combining these two types of weapons, do you think we might see more integrated maneuvers between different parts of the military in the future?
MW: Absolutely. The way of warfare of NATO is a highly interoperable one. When you ask the question, drones or tanks, it’s both, and they need to operate together, so that we have SOPs (Standard Operating Procedures) where drones work alongside tanks and bring distance to the enemy for these survivable platforms.
The state we found when we started with ARX was that if a tank operator wants to operate a drone, he needs to stop the tank and open the shelf or the gate to put the tail operation out and to control the drone. This is not suitable for highly dynamic warfare. These systems need to be linked together.
How does ARX maintain the flexibility of using modular, off-the-shelf parts while still meeting strict military standards for toughness, security, and reliability in combat situations?
MW: The biggest mistake of the established industry is that it doesn’t have the kind of drone we see in Ukraine, because it was looking for the procurement category and couldn’t find it. Does anybody care what capability this robot brings? You use them to get information, to get things from A to B. The nature of the product is different; it’s not like a car, and you don’t need to build a drone like a vehicle.
To change the philosophy of how we categorize them is something that needs to be changed in our procurement cycle. If you design systems as if they were cars, you will never have 50K or 150K sophisticated robots. We will end up with the old, complex, expensive things.
To build a military grade product with confidence off the shelf is to test it. We know which parts need to be military-racked adapters, and which ones are fine to have an automotive adapter, which is 10% of the price of a military adapter, and how to secure this automotive adapter to be rugged enough to be on the battlefield. When we are talking about reusable systems, or attritable systems, it means they need to be cheap enough to lose them, and you cannot go for the 100% solution.
Beyond unmanned ground vehicles, what are the most urgent technology gaps Europe faces in the broader defence tech landscape, such as AI, communications, sensor tech, or space capabilities?
MW: Because unmanned and defence systems rely on interconnectivity, battery technology is an area where Europe lacks sovereignty. Next, space is the most underestimated dimension when it comes to defence, and Europe needs to build its capabilities here as well.
How will soldiers’ day-to-day jobs change with these new technologies? What new skills or training will soldiers need to operate and interact effectively with advanced robotic systems?
MW: Soldiers only had one job: to fight. And this is what they are still doing. But warfare is changing rapidly. These new systems bring distance to the battlefield between the combatants, and for the soldier, it means he or she will find themselves less often in a direct fight with the enemy, and more often as an orchestrator of systems. So it will become more technological.
Of course, it doesn’t mean that it’s safer. It’s just changed, and adaptability is the differentiation between a good army and a bad one, to be honest. For example, the gray zone in Ukraine is shaped by the battery range of unmanned systems, and you need to adapt not only your technology, but also the SOPs (ed. note. Standard Operating Procedures that guide army officers on appropriate actions and protocols) the military forces use: how they work, act, do logistics, and train their soldiers. It needs to change as fast as technology.