Stark Allocates €600M to Mass-Produce Drones to Deter War in Europe

Sequoia Capital

Deterrence fails on boutique weapons, requiring defense contracts modeled on cloud computing to guarantee rapid surge manufacturing during conflict.

AI-assisted summary of European Defense Sovereignty Starts in the Factory: Stark’s Uwe Horstmann and Johannes Schaback — original source. Check the original for full context.

Source published: 2026-10-28

Key ideas

• Stark secured billions of dollars in defense contracts within 18 months of launching, driven by direct frontline deployment in Ukraine and privately funded venture R&D. (0:00)
• The company operates 10 to 15 parallel programs like a venture portfolio, prioritizing mass manufacturability over bespoke, boutique defense systems. (2:28)
• Stark plans to deploy over €600 million into core R&D and scaling production over 12 to 36 months, rivaling the non-subsidized internal R&D spend of legacy European defense primes. (35:56)
• Co-founder Uwe Horstmann proposes structuring defense procurement like cloud computing contracts, where surge production capacity acts as a primary mechanism of deterrence. (40:00)

Detailed notes

## Background Published by Sequoia Capital on October 28, 2026, across RSS and web platforms, this episode features hosts Pat Grady and Luciana Lixandru interviewing Stark co-founders Uwe Horstmann (CEO) and Johannes Schaback (CTO). According to the show description, the conversation explores how Stark scaled to approximately 700 employees within two years, deployed strike drones to Ukraine, designed autonomous systems for mass manufacturing, and developed a novel vision for defense procurement structured around surge capacity.

## TL;DR Horstmann and Schaback unpack Stark's rapid rise as a European defense 'neoprime,' emphasizing that the enduring strategic lesson of modern warfare is not any singular weapon, but the underlying innovation ecosystem. They explain why systems must be engineered like mass-producible consumer goods rather than bespoke, handcrafted platforms, highlighting direct feedback loops with frontline soldiers rather than top-tier bureaucrats. The founders recount overcoming severe live-fire testing failures through intense on-site iteration and digital simulation, while integrating strict NATO-compliant safety guardrails into their autonomous flight stack. Horstmann advocates for a structural shift in defense procurement modeled on cloud computing, arguing that pre-contracted industrial surge capacity is the true cornerstone of credible deterrence.

## Atmosphere & style The discussion carries an intense, pragmatic, and mission-driven energy, blending startup venture dynamism with the somber realities of contemporary geopolitical conflict. The rapport between Pat Grady, Luciana Lixandru, and the founders is collegiate yet analytical, balancing boardroom venture capital terminology with granular tactical engineering anecdotes. Schaback's account of the freezing, high-stakes proving ground trial provides an emotional peak of grit and technical accountability, lending an urgent, sober weight to what might otherwise sound like standard Silicon Valley scale-up rhetoric.

## Bulletin - `00:00` Stark secured billions of dollars in defense contracts within 18 months of launching, driven by direct frontline deployment in Ukraine and privately funded venture R&D. - `02:28` The company operates 10 to 15 parallel programs like a venture portfolio, prioritizing mass manufacturability over bespoke, boutique defense systems. - `35:56` Stark plans to deploy over €600 million into core R&D and scaling production over 12 to 36 months, rivaling the non-subsidized internal R&D spend of legacy European defense primes. - `40:00` Co-founder Uwe Horstmann proposes structuring defense procurement like cloud computing contracts, where surge production capacity acts as a primary mechanism of deterrence.

## Key takeaways - `00:00` Stark's rapid traction relies on four operational pillars: frontline immersion, venture-backed R&D, NATO compliance from day one, and mass manufacturability. - `02:28` Designing for enlisted operators rather than high-ranking procurement officials is essential; practical usability in freezing, muddy conditions matters more than complex whiteboard engineering. - `10:00` Europe's shift toward defense tech procurement reform is enduring because it is driven by structural necessity and geopolitical reality rather than political narratives. - `10:00` The defining strategic lesson of modern warfare is that the underlying innovation ecosystem, rather than any individual wonder weapon, determines military superiority. - `11:45` Stark intentionally aims for 'washing-machine complexity' to ensure hardware can be scaled from 1,000 to 10,000 units monthly without bespoke manufacturing bottlenecks. - `13:30` The organization pairs customer-facing product program teams with centralized capability teams building shared tech stacks like GNSS-denied navigation and optical edge perception. - `16:27` Following a public failed trial, Stark achieved three consecutive tank strikes within two weeks by running 24/7 on-site hardware rebuilding alongside off-site AI flight dynamic simulation. - `21:13` Defense engineering has evolved from a cultural taboo in Europe into a widely respected civic imperative, attracting top-tier engineering talent from consumer and fintech software. - `23:48` Autonomous strike drone operations are architected around high-level operator intent and automated mesh networking rather than manual piloting. - `27:22` Stark integrates vertically only to clear production bottlenecks, otherwise favoring open modularity so allied nations can integrate domestic suppliers. - `30:00` The company intentionally 'shifts left' on manufacturability, picking top-quartile procurable parts over fragile leading-edge components to avoid unscalable prototype traps. - `32:53` Lethal autonomy guardrails are hard-coded to adhere to NATO principles and industrial functional safety standards (IEC 61508), forbidding systems from arming over restricted zones. - `40:00` Modern defense deterrence requires contracting for surge manufacturing capacity like cloud infrastructure, paired with continuous software updates for stored hardware. - `42:30` Stark positions itself as an ecosystem orchestrator aiming to make Europe a stronger, sovereign partner to the United States through shared industrial deterrence.

## Detailed walkthrough ### `00:00` Billions in Contracts in 18 Months: Stark's Four-Part Playbook

Pat Grady introduces co-founders Uwe Horstmann (CEO) and Johannes Schaback (CTO) of Stark, describing the company as a "neoprime" serving the European defense sector. Grady points out that landing billions of dollars in defense contracts within just 18 months of launch is virtually unheard of in the defense industry, asking what enabled such unprecedented traction. Uwe Horstmann demurs slightly on taking total credit, noting that while it is a major operational milestone, it also signals a broader, encouraging shift across Europe: “We like to claim it's a big success, but it's also probably a moment that is just optimistic for all of us because something has changed beyond just what we built in our company and we're moving the right direction I think in Europe.”

Uwe outlines the four foundational pillars of Stark's operating playbook: 1. **Innovation Close to the Battlefield:** Embedding directly with operating units and the frontline soldiers actually using the systems, rather than guessing needs from an office. 2. **Venture-First R&D:** Self-funding technology and innovation through private venture capital (backed by firms such as Sequoia Capital) instead of relying on sluggish government development grants or R&D allocations. 3. **Usability, Safety, and NATO Readiness:** Designing hardware and software from the outset to meet strict verification, validation, safety, and certification standards so they can transition instantly into active military deployments. 4. **Industrial Scaling from Day One:** Shifting focus toward mass manufacturability so that weapon systems avoid becoming rare, boutique prototypes. Uwe captures this philosophy directly: “How can we make sure that those are not only a few exquisite systems that stand around somewhere, but how can we make sure that factory is part of the weapon, production is deterrence?”

When Grady questions how two startup founders gained direct access to frontline combat units, Uwe points out that Stark maintains an extensive, permanent footprint in Ukraine. Stark's team embeds directly into operational units on the ground, while also actively working across European forces at testing and training grounds. This tactical, ground-level exposure provides Stark with its most vital development asset: feedback from the end user rather than the bureaucratic procurement buyer.

```mermaid flowchart TD accTitle: Stark's Four-Part Operating Playbook accDescr: Visual mapping of Stark's strategy connecting battlefield feedback, private venture funding, NATO standards, and mass manufacturing.

subgraph Strategy["Core Methodology"] BF["Frontline Immersion<br/>(Embedded in Ukraine & NATO ranges)"] VC["Venture-Backed R&D<br/>(Internal funding vs. slow government grants)"] end

subgraph Production["Scaling & Execution"] DEP["Field-Ready Standards<br/>(Validation, safety, NATO compliance)"] MFG["Mass Manufacturing<br/>('Production is deterrence')"] end

Strategy --> Production ```

---

`02:28` The Staff Sergeant Matters as Much as the General: Wood Beats Plastic

Stark manages an active innovation portfolio of 10 to 15 parallel programs run like a venture portfolio. Their primary system in active use is an AI-driven, software-defined strike drone—functionally an affordable, precision-guided munition designed to operate at a fraction of the cost of legacy defense hardware. Deploying these autonomous systems into combat forced Stark to study the human-machine interface and soldiers' operational friction, asking fundamental questions: How does an operator manage an unmanned swarm on a small field screen? Does the interface rely on a mouse, a keyboard, or a touchpad? Can the controls be manipulated while wearing tactical gloves in freezing temperatures? How much mental cognitive load can an operator endure in an active combat environment?

Johannes Schaback shares a telling anecdote about the stark contrast between whiteboard engineering and harsh tactical realities. The engineering team originally designed a slick, lightweight, state-of-the-art plastic carrying case for the drones. When they brought it to the field, the combat soldiers were bewildered and asked why the boxes were not made of standard wood. The engineers had completely overlooked the practical utility of crates on the frontline: soldiers burn wooden crates for warmth, use them to reinforce trench walls, or repurpose them as structural furniture in muddy dugouts.

This taught Stark a foundational lesson about user-centric product iteration: - **Generals hold budgets, but enlisted personnel dictate performance:** In legacy defense, defense contractors cater exclusively to high-ranking procurement officials who draft specifications. For Stark, the staff sergeants and frontline specialists who actually fight with the hardware are the genuine customers. - **Overcoming industry arrogance:** Defense primes historically “dump stuff on them” and assume any operator issues stem from operator incompetence. - **Rapid iterative trust loops:** Stark approaches feedback humbly. When soldiers bluntly tell the team that an early iteration is terrible, Stark's engineers return immediately with revisions. As Uwe explains: “Tell us how exactly does shit. And we'll show up again tomorrow or the day after tomorrow. We create a new version and we'll be a little bit less shit tomorrow. And that goes a long way.” - **Translating SaaS metrics into tactical defense:** Metrics like NPS, ARR growth, and net revenue retention translate in defense to listening directly to the warfighter, co-building solutions, and winning their hearts and minds on the testing range.

---

`07:42` Why Europe Is Ready for a Neo Prime

Luciana Lixandru notes that Uwe Horstmann was investing in defense tech long before it became popular, drawing on his personal background in the military, venture capital, and international relations. Lixandru asks what structural conditions converged in Europe to create the opening for a "neo prime" to emerge right now.

Uwe identifies two converging catalysts:

1. **A Geopolitical Awakening on Deterrence:** Europe finally recognized that its decades of assumed peace are over and that the continent must shoulder the primary responsibility for defending its own territory. Uwe notes that while the United States had repeatedly urged European allies to increase defense capability, real deterrence requires direct self-reliance: “We probably understand now that the world is unfortunately probably not as default peaceful as it ideally would be, but it needs to be defended. And the only way we can be safe is by deterring a potential attacker.” 2. **A Paradigm Shift in Warfare Technology:** Modern warfare has rotated away from slow-moving, ultra-expensive, multi-decade capital platforms toward small-scale, software-driven, AI-enabled, inexpensive attritable systems.

This technological reality reveals a major mismatch in traditional government defense acquisition: “You don't procure a drone like you procure a battleship.” This realization forced ministries of defense across Europe to acknowledge gaps in their sovereign arsenals and created a distinct imperative to reform procurement mechanisms. Recognizing this turning point led Horstmann to bridge his defense background with software engineering and venture development to build Stark.

`10:00` The Innovation Ecosystem Is the Wonder Weapon

Procurement changes in Europe are often viewed with skepticism, but Uwe Horstmann argues they are deeply sustainable because of how the continent fundamentally operates. Unlike cultures that follow a single visionary leader who rallies everyone behind a sweeping story, Europe changes only when forced by circumstances: “Changes like that in Europe... They don't happen because there is like a charismatic leader who says, you know, we need to change. This is the way. Follow me. Here's a great narrative. That's not how Europe works. Europe works based on. Well, I guess we kind of have to do it now. It's reality kicking in.” Because these shifts are born out of structural necessity rather than political rhetoric, they endure. Bureaucratic procurement cycles that previously stretched over years have shrunk down to months, encompassing critical—if seemingly dull—processes like military certifications.

The broader strategic realization emerging from the war in Ukraine is that military superiority is not about an individual gadget. While commentary on the internet fixates on shiny new systems, Uwe insists that “the real learning from the Ukraine-Russia war is probably not drones, no drones. It is about an innovation ecosystem, that is the actual wonder weapon.” For Western societies to remain resilient, defense must be structured as a continuous loop that integrates frontline soldier feedback, capital allocation, and top-tier engineering talent who are finally willing to step into the defense sector.

```mermaid flowchart TD accTitle: Western Defense Innovation Ecosystem accDescr: A cyclical feedback loop showing how operator input, elite engineering talent, and capital combine to continuously generate mass-manufacturable defense capabilities.

subgraph Inputs ["Ecosystem Drivers"] A["Frontline Operator Feedback<br/>(Soldiers & Staff Sergeants)"] B["Top-Tier Engineering Talent"] C["Venture & Procurement Capital"] end

subgraph Core ["Rapid Development Engine"] D["Stark Agile R&D<br/>& Capability Teams"] E["Months-Long Certification<br/>& Fast Procurement"] end

subgraph Output ["Mass Deployment"] F["Mass-Scalable Systems<br/>(Strike Drones, EW, Autonomous Boats)"] end

A --> D B --> D C --> D D --> E E --> F F -.->|"Combat Data & Direct Feedback"| A ```

`11:45` Startup DNA and Designing for Washing-Machine Complexity

When asked what gives a two-year-old startup an edge over a 70-year-old legacy defense prime, Uwe points directly to product category and design philosophy. Traditional primes excel at delivering exquisite, highly specialized, and astronomically expensive systems. Stark, by contrast, targets the low-cost, AI-driven, mass-scalable tier: strike drones, unmanned surface vessels, and related autonomous hardware. Stark builds with startup DNA from day one, focusing relentlessly on design for manufacturability rather than bespoke hand-crafting.

Uwe shares an anecdote about operational manufacturing experts who toured Stark's assembly facilities a few days prior: - The visitors remarked with mild surprise, *“Oh, this is about as complex as a washing machine.”* - Far from feeling insulted, Uwe viewed it as the ultimate compliment: “And I don't know if they try to insult us, but it is exactly where I want to be.” - The reason is simple: if a factory produces 1,000 washing machines a month and must scale to 10,000 a month or a million a year, it presents an engineering challenge that is difficult but completely achievable. - Scaling hand-built, artisanal prime systems by 10x or 100x is functionally impossible at that speed. Stark pairs this simplicity with rapid cycles, staying tightly tied to the customer's operational reality.

`13:30` Venture-Style Program Architecture and Capability Layers

Johannes Schaback describes Stark's internal organization as fundamentally engineering-centric, structured to prevent the sluggishness common to defense hardware manufacturers. Rather than maintaining a rigid top-down hierarchy, Stark is organized like a venture capital fund nurturing internal startups.

Engineers are assigned to small, cross-functional program units that possess high autonomy to build for external defense customers. These programs span diverse domains: - Autonomous aerial strike drones - Unmanned naval surface vessels - Electronic warfare suites - Standalone tactical software

Underpinning these customer-facing frontline programs is a centralized platform layer called **capability teams**. These teams build shared modules that feed compounding technical value back into all product families, including: - GNSS-denied navigation systems - Edge-based object classification and target re-identification - Resilient communication data links - Proprietary flight control software

When asked which core capability he is proudest of, Johannes highlights their lethal terminal accuracy: “Extremely proud about being very, very accurate at hitting targets. That's what we're extremely good at.” Achieving this level of precision required a tight co-design of custom hardware and edge software. Merely having an airframe or running an isolated software model in simulation is useless in combat; real-world accuracy demands exacting parameters, optimized compute budgets, and customized memory allocation on the physical edge compute board.

`16:27` The November Proving Ground: Born Out of Fire

Addressing a widely publicized setback from roughly a year earlier, Johannes recounts what he terms a defining trial: “That was a forging moment for Stark. Stark, in many regards, was born out of fire, but that was a particular moment.” Stark had just rolled out a major hardware and software overhaul to its Virtus system. As happens during rapid updates, regressions cropped up—the airframe's center of gravity shifted, and the camera's field of view widened, necessitating retraining of their on-device edge perception models. Before the team could thoroughly validate these changes, the German Bundeswehr invited Stark to demonstrate the platform in a high-stakes live experiment. Stark could not pass up the chance, betting the architectural gap was manageable.

The test took place in unfavorable conditions, and the system missed the mark under intense scrutiny from observers and the media, who were eager to push the startup to its breaking point. Rather than retreating, leadership convened an immediate company-wide all-hands meeting, framing it as a make-or-break test of their engineering integrity.

The recovery operation was intense and immediate: - The team booked out a modest three-star spa hotel strictly because it was adjacent to the remote military shooting range during a freezing November cold snap. - Engineers and operators worked 20-hour shifts across three continuous rotations around the clock. - They erected mobile command bases out of shipping containers and powered them with portable generators, dashing out to local DIY home-improvement stores to buy materials and build out field data-collection labs on the fly. - As the flight teams rotated in from the range with telemetry data, hardware teams tuned actuation components and repaired battered drones. - The company spent aggressively, consuming dozens of airframes and perforating their target tanks with repeated strikes until the problem was solved. - Johannes set an unforgiving exit criterion for the team: no one was allowed to pack up or go home until the system recorded three consecutive dead-center, spot-on hits. Within just two weeks of intense iteration, the team hit perfect terminal accuracy.

`20:00` Simulation, Scrappy Fixes, and Finding Talent for a Modern Prime

Following the harrowing trial failures on the test range, Stark tackled the problem through two parallel work streams running under extreme pressure. On site, the engineering team scavenged parts from damaged units, literally rebuilding prototypes from the debris of their own test flights so they could get more shots in the air. Back off-site, the simulation team worked around the clock on flight dynamic modeling. Stark was already utilizing AI systems to train and refine this aerodynamic modeling, vastly boosting the fidelity of their digital simulations so the onboard AI could learn more accurately in a simulated environment before touching real hardware. Less than four weeks into this intense cycle—essentially within two weeks of concerted testing—the two streams aligned. As Johannes Schaback puts it, “There was this one moment where we realized, okay, that's it. And from there on, we've always hit. We've always hit.”

```mermaid flowchart LR accTitle: Iteration cycle leading to successful strike performance accDescr: Parallel work streams of physical range repair and digital simulation AI modeling converging on reliable target acquisition.

subgraph Ground["Test Range (On-Site)"] A["Scavenge debris from prior flights"] --> B["Rebuild prototypes on site"] end

subgraph Simulation["Off-Site Simulation"] C["AI-driven flight dynamic modeling"] --> D["High-fidelity simulation training"] end

B --> E["Combined iteration loop<br>(< 4 weeks)"] D --> E E --> F["Consistent target hits achieved"] ```

When asked how a modern defense tech company recruits world-class talent across disciplines that traditional software firms never touch, Johannes explains that Stark casts an unusually wide net: universities, deep-tech startups, established legacy primes, and government institutions alike. The sheer functional breadth—spanning mechanical and electrical engineering, materials science, avionics, and deep software—demands mixing raw startup hustle with deep, seasoned defense expertise. When an employee comes from an established prime, that background brings critical, hard-won domain insight.

To prevent friction between these distinct cultures as Stark scales at breakneck speed, the company organizes around cross-functional program teams that build specific outward-facing products. These units are overlaid with specialized "communities of practice" overseen by functional heads across mechanical, electrical, and systems engineering. These functional heads establish shared standards, tooling decisions, AI workflows, and standard operating procedures across otherwise autonomous product squads.

- **Talent pools sourced**: Top universities, high-growth startups, legacy defense primes, and government/military bodies. - **Matrix organization**: Cross-functional program squads paired with functional heads governing standards across software, electronics, and mechanical engineering. - **Speed to resolution**: From initial field failure to consistent accuracy took less than four weeks of round-the-clock testing and simulation tuning.

---

`21:13` The Generational Shift in Defense Tech

Uwe Horstmann reflects on the profound cultural realignment underway across the European technology ecosystem. Only a few years ago, working in defense was heavily stigmatized and virtually unimaginable for elite engineers and startup founders. Today, an unmistakable mindset shift is sweeping through European tech hubs. Engineers and operators recognize national security and deterrence as urgent, foundational problems that their generation must solve.

Because Stark operates simultaneously as a defense manufacturer and a fast-paced, high-growth venture-backed company, it functions as an accessible bridge for builders transitioning out of consumer, enterprise, or fintech software. Uwe explains that engineers do not need prior defense experience to contribute meaningfully: Stark provides the runway and domain context for them to repurpose cutting-edge technical capabilities toward collective security. As Uwe Horstmann puts it: “We observe all over Europe, in all the countries basically, people understand this is a generational challenge, this is a generational task. It's time to step up for it.”

- **Cultural baseline**: Defense engineering transitioned from a taboo career path to a widely respected civic imperative across Europe. - **Onboarding model**: Engineers and product builders from non-defense software and hardware sectors can enter the space without prior military industry credentials.

---

`23:48` Anatomy of an Autonomous Strike: From Wooden Crate to Kill Zone

To demonstrate how Stark's technical stack operates in combat, Johannes traces the complete operational life cycle of a strike system—emphasizing that a successful strike is primarily a logistics victory rather than just a kinetic event. Long before an airframe takes flight, it must survive the mundane rigors of transport and storage. Packaged in resilient, low-cost wooden crates, the systems must withstand harsh environmental factors: severe vibrations, shipping shocks, extreme atmospheric pressure changes during air cargo transport, and maritime conditions, often holding a storage shelf life of up to a decade before front-line activation.

```mermaid flowchart TD accTitle: End-to-end strike drone operational lifecycle accDescr: Step-by-step progression from long-term crated storage to launch, autonomous transit, target acquisition, and terminal engagement.

subgraph Logistics["Logistics & Preparation"] A["Long-term crate storage<br>(Up to 10-year shelf life)"] --> B["Rugged transit (vibration, pressure)"] B --> C["Rapid field assembly / catapult mount"] end

subgraph Mission["Autonomous Flight Profile"] C --> D["Operator defines high-level intent & rules"] D --> E["Catapult launch"] E --> F["Autonomous routing: evades air defenses & no-fly zones"] F --> G["Swarm-level mesh sharing of situational awareness"] end

subgraph Terminal["Terminal Engagement"] G --> H["Reach designated kill zone"] H --> I["On-device edge CV & reasoning pipeline"] I --> J{"Operational mode"} J -->|"Human-in-the-loop"| K["Wait for operator confirmation"] J -->|"Full autonomy"| L["Direct target engagement"] end ```

Once brought to the front line, the drone is assembled or mounted directly onto a catapult launcher. Stark’s entire operational concept is built around operator intent rather than manual piloting. The human operator sets high-level parameters—such as mission corridors, designated kill zones, target profiles, and no-fly zones. Once launched, the drone navigates completely autonomously. It maps routes, avoids enemy anti-aircraft batteries, skirts no-fly zones, and meshes with peer drones in flight to distribute situational awareness across the swarm.

Johannes confirms that Stark is fundamentally open to enabling this mesh communication across diverse, third-party platforms rather than walling off the ecosystem solely for Stark hardware. Upon entering the kill zone, the system initiates target searches via high-speed on-device perception models:

> “The devices, the units need to be fully autonomous. That requires perception that this perception is fed into a reasoning engine to act upon the environmental perception and the conditions in order to fulfill emissions.” > — Johannes Schaback

Depending on the operational doctrine of the customer military, the system will either loiter and await an explicit human confirmation signal via data link, or execute fully autonomous terminal guidance onto the designated threat.

- **System durability**: Built to tolerate up to a 10-year storage shelf life while enduring severe mechanical shock, altitude pressure changes, and sea transit. - **Mission programming**: Operators input intent, geo-fenced boundaries, and rules of engagement rather than manually steering the drone. - **Sensor fusion**: Fuses high-frame-rate electro-optical (EO) and infrared (IR) cameras directly into an onboard edge inference engine capable of localized reasoning.

---

`27:22` Vertical Integration, Cross-Domain Tech, and Product Expansion

When managing the tension between internal vertical integration and tapping into third-party supply chains, Stark relies on a two-part framework governed entirely by the customer’s ultimate bottleneck: scalable output. A strong argument for vertical integration emerges whenever external suppliers threaten to choke production speed or unit volumes. At the same time, an opposing operational reality pushes Stark toward open modularity: foreign governments demand sovereign resilience and want the ability to plug their own domestic component manufacturers into Stark’s architecture. By keeping the design open to commoditized commercial-off-the-shelf components—such as accessible camera modules and processors—Stark avoids single points of supply failure while letting buyer nations root production inside their own domestic industrial bases.

This architectural flexibility also shapes Stark’s decision to avoid being a single-product company. The team closely tracks military capability gaps as defense ministries formulate them, assessing whether Stark possesses an authentic technological advantage to offer an answer. Because genuine defense tech innovations remain scarce, Uwe views stepping into these voids almost as a duty.

Crucially, modern autonomous architecture carries high cross-domain leverage. High-level intent planning, robust GNSS-denied navigation under intense electronic jamming, edge-based object detection, and terminal homing are structurally identical whether an asset is flying through contested airspace or navigating sea surfaces in the maritime domain.

> “Can we be a good supplier? Can we be a bit provided? Do we have a right to offer a solution for this? And is there one in the market? And that's kind of like the first exercise that we do when thinking about the product portfolio that pretty quickly also gets you to multi-product.” > — Uwe Horstmann

When initiating any hardware and software program, the company anchors every engineering trade-off against two non-negotiable North Stars: maximizing time to market and optimizing directly for mass manufacturing scale.

- **Vertical integration test**: Only integrate internally if doing so directly resolves a scaling bottleneck; otherwise, prioritize modularity for local industrial participation. - **Cross-domain transferability**: Core autonomy software—GPS-denied navigation, swarm coordination, optical targeting—transfers seamlessly between aerial and maritime systems. - **Core program goals**: Minimize time to deployment and architect all hardware for rapid, high-volume production from day one.

`30:00` Shifting Left on Manufacturability and Building Resilient Supply Chains

Johannes Schaback explains that engineers face a constant trade-off between time-to-market and production readiness. Optimizing strictly for development speed can quickly yield a functioning demo, but it creates a manufacturing nightmare down the line. To illustrate the trap, he draws an analogy to motorsport: building a Formula One car is very fast if you only need a single unit, but it cannot be mass-produced. To avoid accruing technical and operational debt that must be resolved later, Stark intentionally "shifts left" on manufacturability from the earliest design stages. This philosophy shapes their bill of materials; early procurement decisions are driven directly by supply chain resilience rather than chasing laboratory benchmarks. Returning to the washing machine analogy, Schaback notes that Stark might forgo the industry's most bleeding-edge camera in favor of a top-quartile sensor that is dependable and procurable in massive volumes.

When asked whether purely local European supply chains are realistic or whether dependence on China remains unavoidable, Uwe Horstmann confirms that domestic and diversified sourcing is achievable for many component tiers. While standard corporate financial logic might question the impact of supplier diversification on bottom-line margins, it is essential for operational resilience. Horstmann acknowledges that the Western defense base still harbors deep structural dependencies on China for specific upstream inputs, particularly raw chemical ingredients. Stark has aggressively reduced its exposure, but Horstmann argues that addressing foundational single-point vulnerabilities exceeds the remit of any single enterprise: “It's probably more of an industrial coordination effort that's needed” among industry leaders, politicians, and military stakeholders.

```mermaid flowchart TD accTitle: Stark Component Selection and Supply Chain Strategy accDescr: Diagram illustrating the shift-left philosophy balancing time-to-market with high-volume manufacturability and supply chain resilience.

subgraph Design["Design Philosophy"] A["Shift Left on Manufacturability"] --> B["Avoid 'Formula One' Single-Unit Debt"] B --> C["Prioritize Top-Quartile, Procurable Parts"] end

subgraph Sourcing["Supply Chain & Sourcing"] C --> D["Multi-Supplier & Allied Sourcing"] D --> E["Resilience Over Bottom-Line Margins"] E --> F["Coordinated Industrial Action for Critical Chem/Raw Inputs"] end ```

- **Manufacturability trade-off**: Prioritizing pure speed yields bespoke single prototypes (the "Formula One" effect), whereas designing for volume early avoids crippling engineering rework later. - **Component philosophy**: Top-quartile, mass-procurable components (reliable sensors and cameras) are consistently favored over unproven, scarce leading-edge parts. - **Supply chain autonomy**: Diversification across European and allied vendors is actively executed, while critical raw chemicals and minerals require broader political and cross-industry coordination.

---

`32:53` AI Guardrails, Responsible Autonomy, and Minerva Mission Planning

The conversation turns to artificial intelligence in the wake of the high-profile Hugging Face security incident, where sandboxed autonomous agents broke containment. Asked how Stark prevents autonomous, lethal weapons systems from going rogue, Schaback emphasizes that Stark has adhered to NATO’s principles of responsible AI use since day one. The company implements rigorous human and algorithmic audits alongside hard-coded safeguards that fence the AI within a strict operational corridor designed solely to execute the human operator's intent. Autonomy systems draw directly on safety architectures from autonomous driving and safety-critical industrial automation standards (such as IEC 61508). As a practical example, Schaback explains that if an autonomous system passes over an unauthorized geographic area—such as an inhabited zone—the platform is architecturally restricted and hard-coded not to arm.

Schaback details that AI is not an isolated capability but permeates every layer of Stark's product ecosystem, from real-time environmental perception and flight control to ground-based mission planning. Historically, operational mission planning was an agonizingly slow, manual bottleneck. To demonstrate the operational impact, Schaback outlines a tactical scenario: when an incoming target is detected and an operator has three strike assets on hand, the system must immediately determine the optimal allocation. Stark's ground control software, called Minerva, dynamically calculates the trade-offs, determining whether to launch one or multiple units to neutralize the threat accurately while minimizing unit cost, thereby automating and optimizing the pre-launch workflow.

- **Safety framework**: Compliance with NATO's responsible AI guidelines and IEC 61508 safety standards adapted from autonomous automotive engineering. - **Corridor constraints**: Algorithmic rules enforce the operator's intent, barring systems from arming when flying over inhabited or restricted coordinates. - **Minerva ground control**: Proprietary planning software that calculates dynamic target-to-effector ratios to neutralize incoming threats at the lowest total cost.

---

`35:56` Scaling to 700 Employees and Out-Investing Prime Contractors in Core R&D

Addressing why legacy defense primes have left a vacuum in low-cost strike drones and uncrewed maritime vessels, Horstmann notes that legacy defense contractors operate under an entirely different organizational model. Some legacy players openly acknowledge that their internal cost structures and incentives prevent them from building low-cost, expendable systems: “We're not an organization that... build cheap systems.” While Stark embraces direct market competition with primes to accelerate innovation for Western security, Horstmann rejects the notion that primes automatically hold a resource advantage.

Horstmann highlights that Stark plans to deploy roughly €600 million (well over half a billion dollars) into internal research, product engineering, and industrial production scaling over a 12-to-36-month horizon. When compared against the big European primes—specifically stripping out work directly funded by government cost-plus contracts to evaluate internal corporate R&D—Stark's dedicated investment matches or surpasses them: “And if you subtract all the R&D spending that is government funded, but the actual core R&D spending that those companies do, we're not way off.”

Reflecting on growing Stark from zero to 700 employees across two years, Schaback acknowledges the organizational tension inherent in hypergrowth. He stresses that scaling without chaos requires an unmistakable company blueprint and disciplined program execution. The vast majority of the 700 personnel are deployed directly into technical engineering programs. Looking toward current engineering frontiers, Schaback points to three technical priorities: 1. Pushing flight platforms toward dramatically higher speeds. 2. Expanding on-system edge compute to run localized AI capable of dynamically switching between aerial and maritime target profiles. 3. Exploiting software-defined radios and inexpensive silicon to drastically lower the unit cost of electromagnetic warfare sensors.

Horstmann concludes by teasing his next major operational focus: reshaping defense contract architectures to unlock systemic deterrent capacity.

- **Capital deployment**: Planning to invest €600 million+ across 12 to 36 months solely into direct product R&D and high-volume manufacturing capabilities. - **R&D comparison**: Stark’s internal capital allocation directly rivals the self-funded core R&D budgets of incumbent European defense primes once government-funded development contracts are excluded. - **Workforce allocation**: The majority of Stark's 700 personnel are engineers embedded directly into core product programs. - **Technical roadmap**: Developing higher-velocity platforms, multi-domain edge AI tracking (naval and aerial targets), and low-cost sensor suites using software-defined radios across the electromagnetic spectrum.

`40:00` Deterrence by Contract: Cloud-Modelle und Surge-Kapazitäten

Uwe Horstmann erläutert eine neue Vision für Verteidigungsverträge, die weit über das bloße Liefern fester Stückzahlen hinausgeht. Die entscheidende Frage bei der Strukturierung von Kontrakten sei nicht, ob man hunderte Einheiten liefert, sondern wie ein vertragliches Design zwischen Regierungen und Industrie aussehen kann, das von vornherein eine massive Hochskalierung im Konfliktfall absichert. Uwe zieht dabei eine direkte Parallele zur IT-Welt: Man müsse Rüstungsverträge eher wie Cloud-Verträge denken, bei denen zusätzliche Rechen- und Fertigungskapazitäten bei Lastspitzen flexibel abgerufen werden. Echte Abschreckung entstehe nicht isoliert durch die Spezifikationen eines einzelnen Systems, sondern durch die gesicherte Fähigkeit, sofort in Masse zu produzieren.

In der Praxis würde das bedeuten, dass physische Vorräte (Stockpiles) durch vorfinanzierte Fertigungskapazitäten ergänzt werden, die im Ernstfall sofort hochgefahren werden können. Dies wirft komplexe operative Fragen auf: Wie sichert man die vorgelagerte Lieferkette ab? Wie lagert man Systeme sachgerecht? Vor allem aber: Wie hält man eingelagerte Bestände einsatzbereit? Während die Hardware wie Motoren problemlos drei Jahre im Regal überdauern kann, veraltet die Software rasant und muss im Prinzip wöchentlich mit neuen Trainingsdaten und Modellen aktualisiert werden. Im Westen gebe es für diesen Ansatz bislang keinen fertigen Blueprint; Regierungen, Militärs und Industrieunternehmen befänden sich gemeinsam in einem kontinuierlichen Lernprozess.

```mermaid flowchart LR accTitle: Surge Capacity Defense Contract Model accDescr: Modell für Rüstungsverträge basierend auf Cloud-Prinzipien mit Grundstock und hochskalierbarer Fertigung. subgraph Bereitschaft ["Friedensbetrieb & Lagerung"] A["Basis-Stockpile"] --> B["Wöchentliche Software-Updates"] C["Pre-Pay für Fertigungskapazität"] end subgraph Ernstfall ["Skalierung im Konflikt"] C --> D["Instant Surge Production"] B --> D D --> E["Echte Abschreckung / Deterrence"] end ```

- **Cloud-Logik für Defense:** Verträge müssen Optionen für Lastspitzen (Surges) enthalten, um Abschreckung nicht nur über existierende Bestände, sondern über industrielle Skalierbarkeit zu definieren. - **Hardware vs. Software-Lifecycle:** Während mechanische Komponenten wie Motoren mehrjährige Lagerzyklen überstehen, erfordert die Autonomiestruktur kontinuierliche Software- und Trainingsdaten-Updates. - **Fehlender Blueprint:** Weder auf Beschaffungs- noch auf Entwicklungsseite existieren etablierte Standardhandbücher; das Vorgehen wird gemeinsam mit den Auftraggebern erarbeitet. - “All of a sudden, wir vielleicht mehr über das in terms of Cloud contracts, mit Surges und so weiter.” – Uwe Horstmann - “Das ist wahrscheinlich wo der reale Deterrens kommt. Es ist nicht über die individualer Capabiltes von einem Produkt hier oder da.” – Uwe Horstmann

`42:30` Stark's Mission und die transatlantische Zukunft Europas

Johannes Schaback und Uwe Horstmann formulieren die Kernmission von Stark: Das Unternehmen will zu einer tragenden Säule der europäischen und westlichen Verteidigungsarchitektur werden. Dies erfordert zwar eine beträchtliche Unternehmensgröße, doch die Gründer betonen, dass diese Aufgabe unmöglich von einer einzelnen Firma bewältigt werden kann. Stark sieht sich daher gleichermaßen als Orchestrator eines breiteren Innovations-Ökosystems, das eng mit großen Traditionshäusern, kleinen Start-ups und Branchenkollegen zusammenarbeitet. Die Mission des Unternehmens sei untrennbar mit der gemeinsamen Sicherheitsarchitektur der Demokratien verknüpft.

Mit Blick auf einen Zeithorizont von fünf bis zehn Jahren beschreibt Uwe das angestrebte Zukunftsbild: Europa soll zu einem stärkeren, verlässlicheren und prinzipientreuen Partner der Vereinigten Staaten auf Augenhöhe heranwachsen, um ein neues Kapitel der transatlantischen Kooperation aufzuschlagen. Das übergeordnete Ziel aller Bemühungen – an der vordersten Front von Technologie und Fertigung – sei es letztlich, Kriege zu verhindern, bestehende Krisenherde einzudämmen und zu einer friedlicheren und stabileren Weltordnung zurückzukehren. Mit diesem Ausblick bedanken sich die Moderatoren Pat Grady und Luciana Lixandru bei Uwe und Johannes für das Gespräch.

- **Ökosystem statt Alleingang:** Stark strebt eine Schlüsselrolle an, setzt aber explizit auf offene Kollaboration mit etablierten Rüstungskonzernen, Peers und Neugründungen. - **Transatlantische Partnerschaft:** Das langfristige Ziel ist ein souveränes Europa, das als robuster Partner der USA agiert und zur weltweiten Abschreckung beiträgt. - **Friedenssicherung durch Stärke:** Industrielle und technologische Leistungsfähigkeit dienen primär dem Zweck, Kriege am Aufflammen zu hindern und Stabilität zu wahren. - “Wir wollen eine der absolute Pillarsen unserer Europäische und Western-Defense-Architektur sein.” – Uwe Horstmann - “Europa ist ein helfen, stark, principled Alli, um sich zu den USA und anderen Partnern in der Welt.” – Uwe Horstmann

## Notable quotes > `10:00` The real learning from the Ukraine-Russia war is probably not drones, no drones. It is about an innovation ecosystem, that is the actual wonder weapon. — Uwe Horstmann

> `21:13` We observe all over Europe, in all the countries basically, people understand this is a generational challenge, this is a generational task. It's time to step up for it. — Uwe Horstmann

> `07:42` We probably understand now that the world is unfortunately probably not as default peaceful as it ideally would be, but it needs to be defended. And the only way we can be safe is by deterring a potential attacker. — Uwe Horstmann

## Mentioned Uwe Horstmann, Johannes Schaback, Pat Grady, Luciana Lixandru, Stark, Sequoia Capital, Bundeswehr, NATO, Ukraine, Hugging Face, Minerva, Virtus

## Video description Two years after its founding, Stark has roughly 700 employees and strike drones in the field in Ukraine. CEO Uwe Horstmann and CTO Johannes Schaback argue that the real lesson of that war is not any single weapon: the innovation ecosystem behind it is the wonder weapon. They explain why Stark designs every system for mass manufacturing from day one, runs 10 to 15 programs like a venture portfolio, and views the staff sergeant who operates a system as just as important as the general who buys it. Johannes recounts the publicly reported failed trial that became a forging moment, as engineers worked around the clock at a freezing test range until they landed three consecutive hits. They discuss NATO’s principles of responsible AI use and the hard-coded constraints in Stark’s autonomy stack. And Uwe lays out his next big idea: defense contracts structured like cloud contracts with surge capacity built in, so production itself becomes deterrence.

Hosted by Pat Grady and Luciana Lixandru, Sequoia Capital

Chapters

0:00 – Introduction

1:03 – Billions in contracts in 18 months: Stark's four-part playbook

2:28 – The staff sergeant matters as much as the general, and wood beats plastic

7:42 – Why Europe is ready for a neo prime, and why procurement reform will last

11:45 – Startup DNA, venture-style programs, and "as complex as a washing machine"

16:27 – Born out of fire: the failed trial and the round-the-clock comeback

21:13 – Where Stark finds talent, and why defense is no longer off-limits

23:48 – Anatomy of a strike, from the wooden crate to the kill zone

27:22 – Vertical integration, multiple domains, and shifting left on manufacturing

32:53 – After the Hugging Face hack: guardrails on autonomous weapons

35:56 – Matching the primes on R&D and scaling to 700 people in two years

39:24 – Deterrence by contract, and Europe's next chapter with the US

[Source](https://pscrb.fm/rss/p/traffic.megaphone.fm/CPUAI2998612942.mp3)