European Defense Sovereignty Starts in the Factory: Stark’s Uwe Horstmann and Johannes Schaback
In a Nutshell
Stark has achieved unprecedented defense contracts by designing attack drones and autonomous systems for rapid scalability from the factory floor, while embedding engineering teams directly with frontline units to iterate based on actual soldier feedback rather than procurement officials. Their approach treats production capacity itself as a deterrent, building systems with industrial expansion in mind from day one rather than retrofitting prototypes for mass manufacturing. The company operates like an investment fund managing 10-15 independent programs simultaneously, leveraging AI throughout operations while maintaining human oversight per NATO doctrine, and is positioning itself as a cornerstone of European defense by solving the scaling challenges that traditional contractors cannot address.
These notes were generated by AI and may contain inaccuracies.
Europe has reached a clear realization of the need to take greater responsibility for its defense. The United States has confirmed this repeatedly, and Europe must upgrade its defense capabilities as the way to continue living in peace through deterrence.
Uwe Horstmann and Johannes Schaback are co-founders of Stark, a leading defense company serving the European defense community. The company has won billions of dollars in contracts within 18 months of founding, which is unprecedented in the defense industry.
Stark achieved success through four specific approaches:
- Adopting innovation close to the battlefield by fully integrating with units and soldiers who use the systems
- Adopting a pioneering investment approach, funding innovation and product development themselves rather than relying on government spending for research and development
- Designing products for usability and wide deployability beyond mere innovation, including certification and system integrity verification for NATO forces
- Designing products with industrial expansion in mind from the beginning, making the factory an integral part of the weapon and production a deterrent
Stark maintains a large team in Ukraine and integrates intensively with units there. They also work with European forces. The best information comes from interacting with soldiers, being with them, going to the battlefield, and test and training grounds.
The most important input is the customer's opinion, which is not necessarily the buyer's. The buyer is often a high-ranking official, while the customer is often an ordinary soldier at the tactical operations level. This feedback from actual users is the most important thing for the company.
At any given time, Stark manages between 10 and 15 different programs, similar to an investment innovation portfolio. The current program involves attack drones - ammunition that functions more like a weapons system relying heavily on software and artificial intelligence, relatively inexpensive compared to other weapon systems.
The company is still at the beginning of understanding the mental burden on soldiers operating unmanned squadrons. Questions include how the control box should be designed (wood versus plastic), how to visualize a squadron on a small screen, and how a soldier can operate it.
A story illustrates the difference between lab and field conditions: a nice, lightweight plastic box was brought to soldiers who asked what they would do with it, suggesting firewood instead because it could be used for a fire to keep them warm in winter. This feedback had not been anticipated during design.
In military procurement, generals are often considered most important because they control budgets, but non-commissioned officers and soldiers who actually use the systems are more important. The startup spirit requires being close to the customer and developing products for them. The customer includes both the buyer and the user, who are not the same people in military situations.
Questions about swarm drone operation include whether to use a mouse, keyboard, or touch panel, how this works with gloves, what clicks are allowed, where extreme precision is needed, mental workload considerations, and how much information can be processed simultaneously.
Soldiers are used to companies presenting products as "the best" and dismissing problems as user error. Stark takes a different approach by treating soldiers as customers whose time and feedback are most valuable. When told a product is "very bad," they ask for exact details, return with new versions, and iterate quickly. This approach has gained trust in the community.
A favorable opportunity exists in Europe to establish defense companies. Uwe saw this opportunity before others, investing in defense before it became obvious. He has a military background.
Several factors have converged: Europe's realization it must take greater responsibility for defense, awareness of battlefield realities, emergence of new technologies based on software and artificial intelligence with low production cost and small size, and government recognition of the need to fill capability gaps.
Buying a drone is completely different from buying a warship, requiring changes to purchasing mechanisms. This coincided with the establishment of companies willing to meet these new requirements.
Uwe's background combines interest in international military and security affairs, international relations, and technology, creating a clear moment to step into the defense sector.
The changes are sustainable because they are not driven by charismatic leaders but by self-evident truths based on real, fundamental structural changes. Europe operates on the basis of collective recognition that change is necessary. Procurement processes that used to take years are changing to months for certifications.
The real lesson from the Ukrainian-Russian war is not about drones but about an ecosystem of innovation. The true miracle weapon is the ability to continuously innovate in defense, including combat soldier input, talent attraction, and capitalist approaches.
As a two-year-old company, Stark has advantages over 70-year-old companies in low-cost, AI-powered, widely scalable systems like attack drones and unmanned boats. They built prototypes like startups do, from the beginning designed for large-scale manufacturing. When operational experts visited factories, they compared the complexity to a washing machine, which is exactly the goal - systems that can scale from 1,000 units per month to 10,000 per month or a million per year.
Large established companies excel at building expensive, critically important systems and will continue playing a major role. However, rapid innovation, continuous development, close customer communication, and building highly scalable systems represent Stark's strong competitive advantage.
Stark is fundamentally an engineering company. The organizational structure builds programs that can operate and develop very independently. The company operates like an investment fund, building small startups within the organization. Engineers quickly find themselves within a family of highly independent programs working cross-functionally.
Programs include fully autonomous maritime systems, fully autonomous aerial systems, attack drones, electronic warfare, and software.
A platform layer called capability teams builds capabilities that enhance the ecosystem, including navigation without GNSS, object classification, target reassignment, communication links, and flight control. Engineers either work on the platform helping the entire suite of systems or focus on specific clients.
Stark is extremely proud of their high accuracy in hitting targets, requiring a highly integrated approach tightly combining hardware and software systems. Achieving accuracy of less than one meter to hit a tank or moving target at 60 km/h requires precise adjustments in computing, computing budget, and memory allocation.
About a year ago, a major upgrade to the Virtus system including both hardware and software resulted in performance drops. The German army (Bundeswehr) requested participation in an important experiment. The system didn't meet ambitions, and the test environment wasn't ideal. Media sensationalized the situation, creating a defining moment for the company.
An emergency meeting was held, and the best engineering minds worked together in a three-star hotel near the firing range in freezing November conditions. They worked three shifts around the clock. Flight test teams collected and transmitted data while software engineers improved flight tests and perception. Hardware engineers refined the operating system. Within two weeks, they approached perfect accuracy.
The team decided not to leave until achieving three consecutive accurate hits. Tanks were badly damaged from repeated firing, and systems were wasted, but money didn't matter. They worked 20-hour shifts, with flight test teams repairing systems and removing debris for reassembly. The simulation team refined flight dynamics modeling using AI systems to train models.
After less than four weeks, the solution was achieved, and since then the company has always been successful in tests.
Talent is sought from traditional industry, universities, startups, and government institutions. The needed talent comes from everywhere, covering senior to entry-level positions. Established companies bring valuable knowledge, but startup mindset is equally important.
Cross-functional programs develop products with function heads ensuring coherence across electrical engineering, mechanical engineering, and other functions. Communities of practice within cross-functional teams share knowledge, best practices, tool decisions, AI workflows, skills, and MCP certifications.
A cultural shift is happening in the tech and startup community. A few years ago, joining defense was unthinkable, but this is changing across Europe. People recognize this as a generational challenge and task. Stark serves as a good starting point because it functions as both a defense company and fast-growing tech company.
The operational concept starts with logistics - how systems are disassembled, reassembled, transported, and withstand shocks, vibrations, and pressure. The operator prepares the plan structured as no-fly zones, kill zones, or specific mission elements. The system executes autonomously, finding its way, avoiding no-fly zones and anti-aircraft installations.
If part of a squadron, the system communicates via the swarm, exchanging information about what it sees and sharing situational awareness. The company is open to systems working across different manufacturers through communication mechanisms and modularity.
Upon reaching the destruction zone, the system searches for targets. Depending on client requirements, it either searches and destroys targets completely autonomously or waits for operator confirmation before completing the task.
High-speed, high-resolution computer vision technologies process data on-device. The devices must be completely autonomous, requiring perception fed into an inferential engine. Components are sourced from various suppliers including infrared and photoelectric cameras, integrated as sensors for situational awareness.
The war in Ukraine changes daily, requiring flexibility in system design. Components are purchased in bulk and integrated as sensors. The company continuously learns about new types of targets.
The fundamental principle guiding decisions on vertical integration is solving the customer's problem, with scalability being a primary concern. The strong argument for vertical integration is whether it can solve the scalability problem. There's a counterbalancing pull toward integrating different suppliers to make everything more agile and to enable sales to other countries where they can integrate their own suppliers, because production should be rooted in the purchasing countries. This creates an ongoing balancing act between scalability needs and supplier integration for agility.
The approach to product portfolio begins by identifying capability gaps that militaries themselves describe. The evaluation process asks whether the company can be a good supplier, offer meaningful help, and has the right to offer a solution, while checking if similar products exist on the market. This leads quickly to a multi-product approach because there's no abundance of offerings in the areas of operation. There's a sense of obligation to try filling these gaps, while balancing the complexity of multiple programs against the risk of losing focus.
Many technological and operational concept aspects can be applied across different domains. For example, intent planning and mission planning work whether vehicles are flying or swimming. Navigation in restricted environments with interference occurs in both maritime and airspace domains. Object detection and identification applies across domains, as does targeting specific objects. These shared capabilities enable the company to contribute to identifying gaps and helping solve problems across multiple domains.
Two competing goals exist when starting any program: improving time to market and improving production readiness. Focusing only on time to market may result in a product that's quick to demonstrate and test but difficult to manufacture at scale, similar to a Formula 1 car where only one is ever made. From an engineering perspective, manufacturability must be a focus from the beginning to avoid accumulating technical debt. Supply chain flexibility plays a pivotal role from the moment components are selected, leading to choices like selecting very good cameras from the top quarter rather than the absolute best, if they can be purchased in large quantities.
Local supply chains in Europe are considered realistic, particularly for certain components where diversification is relatively straightforward. This is beneficial from both profitability and flexibility perspectives. However, the Western world remains entirely dependent on some components, such as Chinese supplies, and efforts are underway to reduce this dependency significantly. Current systems are described as almost entirely self-sufficient, though some areas require industry-wide coordination with politicians and military to develop strategies, particularly for raw chemical components that may exceed single-company scope.
From the beginning, Stark has applied the NATO doctrine of use, with a person in charge of artificial intelligence. This necessitates scrutiny of systems not only by humans but through layers of safeguards that protect the AI within a limited scope to execute operator intentions. This governs the architecture of full autonomy in every system and every phase of the mission. Pre-programmed constraints depend on location—for example, systems may not be activated when flying over certain populated areas. Additional safeguards include the IEC 61508 standard and elements borrowed from autonomous driving that have been integrated into autonomous systems.
AI appears in every part of the product and every phase of the mission, including decisions about system perception, environmental interaction, and ground-based mission planning before takeoff. Mission planning, which used to be a complex and lengthy process, is being transformed through AI. In practical examples like incoming targets where operators have three systems available, AI helps determine how many systems should be launched and the most cost-effective way to counter threats. The ground control system called Minerva then optimizes mission planning to counter threats accurately and at the lowest cost.
Traditional leading companies are involved in this space to varying degrees. Some have entered the market while others state this isn't their core business, claiming they're not organizations that build cheap systems but focus on different market segments. The key is starting a competitive process where all participants improve together, recognizing that ultimately everyone is on the same team. Competition is welcomed as a path to collective improvement.
Over the next 12, 24, or 36 months, approximately €600 million will be invested in research and development, products, and building production capacity. When comparing this to R&D spending of the biggest European companies (excluding government R&D), there's no huge difference in baseline spending. The focus remains on staying agile while targeting specific market segments and making meaningful contributions through customer understanding, building exactly what's wanted, and gathering feedback from the most junior employees.
From zero to 700 people in two years, growth comes with compromises and difficult decisions, but the process is manageable. The company's plans must be very clear, and while growth is constant, the plan is working well. The critical problem being solved is current and urgent, requiring quick action without the option of slowing down. Most of the 700 people are engineers, which is essential for programs of such critical importance and value.
Systems in Ukraine are evolving at a much faster pace than observed elsewhere. Computing is becoming more accessible, and efforts focus on enhancing artificial intelligence in systems securely while giving greater target-handling capabilities adaptable to maritime and aerial targets beyond just land-based ones. The electromagnetic field is experiencing radical transformation through software-defined radios and affordable computing, enabling cost reduction for electromagnetic sensors.
Contract design is considered the next major challenge with significant potential for deterrence. The focus extends beyond selling hundreds or thousands of systems to structuring contracts between government and industry that include everything necessary for massive research and scaling needed in hot conflict scenarios. This could resemble cloud computing contracts that account for sudden increases in demand. Key elements include reserve stockpiles, prepaid manufacturing capabilities that can be activated immediately, supply chain considerations, system storage, and responsibility for ongoing updates, as drone engines might last three years while software requires weekly updates with new training data.
No one in the Western world has figured out the optimal structure for these arrangements. There's no ready-made plan, and this applies to all procurement and product development work. Everyone—government, military, and industry—is learning together without hundreds of people who have done this work before and can teach others. There's optimism about finding solutions because the right tools don't yet exist, and the principle is that this problem must be solved together.
The mission is to be one of the cornerstones of the European and Western defense architecture—a large company delivering the discussed capabilities while helping create a real innovation ecosystem. The goal includes being good collaborators with large companies, small companies, peers, and everyone, recognizing that one company's mission is closely linked to all missions. The aim is to be a key pillar, indeed a dominant pillar, though this isn't a task one company can accomplish alone.
In 5 or 10 years, Europe should once again become a better partner for the United States, with hopes for a new chapter of transatlantic cooperation. Europe would be a strong, useful, and principled ally among itself and with the United States and other partners worldwide. The vision includes living in more peaceful times, returning to conditions from a few years ago, easing some wars and crises, and ensuring no more wars break out.
Keep Sequoia Capital in your library
Save the videos and channels worth coming back to, and find them again in one place.
