From space capsules to neurosurgery microrobots, deep tech startups require massive capital and long timelines. Is venture capital structurally equipped for this, or are we forcing the wrong model onto the hardest problems? We believe that venture capital is structurally equipped for part of the journey of a deep tech company, but not for all of it. The mistake is to assume that every technology company can be financed with the same model as a software company. Deep tech requires a more thoughtful capital stack: venture capital for the company-building phase, non-dilutive funding for technical risk, strategic partners for validation, and eventually debt or project finance for infrastructure-like scaling.

At APEX, we do not see deep tech as “too hard” for venture. We see it as a different risk curve. In the beginning, the scientific and technical risks are higher. But once those risks are reduced, defensibility can be much stronger than in many traditional tech companies. Patents, know-how, physical infrastructure, regulatory approvals, and deep customer integration create barriers that are difficult to replicate.

So the answer is not that venture is the wrong model. The answer is that venture alone is often incomplete. The best deep tech companies are built with hybrid financing, strong milestone discipline, and investors who understand both the science and the commercial path.

Europe is producing breakthroughs, from satellites like FOREST-3 to AI imaging and quantum research. But we still struggle to build global leaders. Where exactly does execution break down? Europe does not have an ideas problem. We have outstanding research, strong engineering talent, and a growing number of founders who are willing to build companies around very complex technologies. The execution gap appears later: when a scientific breakthrough needs to become a product, when a pilot needs to become a commercial contract, and when a promising company needs to scale internationally.

There are several recurring bottlenecks. The first is speed. European startups often lose time in bureaucracy, fragmented regulation, slow procurement, and complex university technology transfer. The second is capital. We have improved significantly at seed and Series A, but growth capital remains too fragmented and too cautious. Europe does not have a start-up problem; we have a scale-up problem. The third is commercial ambition. Many companies are excellent at proving that a technology works, but less prepared to define a sharp market entry, build a sales organization, and compete globally from day one.

The execution gap is therefore not in the lab. It is in the translation layer between lab, market, customer, and capital. Europe needs more first customers, faster procurement, stronger industrial partnerships, and investors willing to support companies beyond the proof-of-concept stage.

Companies like Contextflow or OroraTech show strong technology, but deep tech often fails at commercialization. What separates companies that scale from those that stay technically impressive? The companies that scale are usually not the ones with the most impressive technology alone. They are the ones that turn technology into a repeatable commercial product. That sounds simple, but in deep tech it is often the hardest step.

What separates the best companies is a very clear understanding of the customer’s pain point. They do not simply ask, “Is our technology better?” They ask, “Which urgent problem does this solve, who pays for it, how does it fit into the existing workflow, and what economic value does it create?” In healthcare AI, for example, technical performance is only one part of the equation. Reimbursement, clinical workflow integration, regulatory approval, and trust from physicians are just as important.

Contextflow is a good example of a company that moved beyond being an impressive AI technology. It built around a concrete clinical use case, integrated into radiology workflows, and created strategic relevance for a larger international player. OroraTech is another example: the real value is not only in launching satellites, but in turning thermal data into actionable wildfire intelligence for customers.

Deep tech companies scale when they stop selling science and start selling outcomes.

With developments like PHOENIX 2 or growing space infrastructure plays, where do you actually see near-term commercial returns in space tech and where is capital still chasing narratives? In space tech, I see near-term commercial returns where space is not the customer, but the infrastructure layer for solving a problem on Earth. Earth observation, wildfire detection, climate monitoring, secure communications, space traffic management, and orbital logistics are areas where the customer need is increasingly tangible.

OroraTech is a strong example. The value is not in the satellite as an object; the value is in near-real-time wildfire intelligence that helps governments, insurers, utilities, and landowners make better decisions. Similarly, companies like OKAPI:Orbits address a very practical problem: as orbit becomes more crowded, satellite operators need automated, reliable ways to manage collision risk and space sustainability.

With ATMOS Space Cargo, orbital return infrastructure is also becoming more relevant. If we want to use microgravity for life sciences, materials, or defense-related applications, we need reliable and affordable ways to bring payloads back to Earth. That is infrastructure, not science fiction.

APEX is focusing on AI infrastructure, quantum and edge computing, where do you see the next inflection point where these technologies stop being ‘future bets’ and start creating real markets? If you had to cut one deep tech trend entirely from your investment focus today, because it’s overhyped or too early, which one would it be? The next inflection point will come when AI infrastructure, quantum, and edge computing stop being separate technology categories and become part of the same industrial stack. AI is creating enormous demand for compute, energy efficiency, data movement, and new architectures. That pressure will accelerate innovation in semiconductors, photonics, edge AI, and eventually quantum computing.

I do not believe quantum will suddenly replace classical computing. The more realistic path is hybrid: quantum systems solving specific classes of problems where classical systems struggle, while AI and high-performance computing handle the broader workload. The first real markets will likely emerge around simulation, optimization, materials, chemistry, and highly specialized industrial applications.

Edge computing is also becoming more important because not all intelligence can sit in centralized data centers. Robotics, autonomous systems, medical devices, industrial automation, and defense applications require low latency, resilience, and local decision-making. That creates opportunities for new hardware, software, and infrastructure layers.

If I had to cut one trend, I would cut generic “quantum software” companies that depend on fault-tolerant universal quantum computers arriving quickly, but do not have a credible commercial wedge before that. We remain very convinced about quantum, but timing and specificity matter. Deep tech investing is not about chasing the most futuristic narrative. It is about identifying the point where a breakthrough starts to become a market.