Scaling Deep Tech Depends on Three Levels of Infrastructure

Every technology thesis eventually collides with the same question: where does it get built? The history of industrial transformation is not primarily a history of ideas but a history of infrastructure. The steam engine required iron foundries, the semiconductor required the cleanroom, and electric vehicles require, right now, a complete reinvention of the battery supply chain, the charging network, and the grid itself. Deep tech ventures that ignore this question do not fail because their science was wrong; they fail because they underestimated the infrastructure gap between a working prototype and a product that can be manufactured, financed, regulated, procured and deployed inside real industrial systems. 

Before deep tech had a name, it came from government laboratories and defence programs with decade-long funding horizons and no obligation to a market, a model that has not disappeared: the EIB Group confirmed more than EUR 2.4 billion in support for energy resilience and deep tech innovation at Hannover Messe 2026, showing that public capital remains central to the long cycles that deep tech requires. 

Alongside this model, a second way of building has emerged, one in which the translation from laboratory to market is treated as a design problem from the outset. Thus, defensibility is built not only through patents, but through deployment infrastructure, operational data, and the learning accumulated as the technology is used in real conditions. 

Through the Nurture Program, Challenge+, and the Creative Destruction Lab, the HEC Deep Tech Center works with founders across this full arc: from the moment a scientist first asks whether their discovery has commercial application, through to the point where a venture is scaling production. What we have learned from working across climate, space, AI, and computing is that the infrastructure challenge has a consistent structure, regardless of the technology domain, one that plays out at three scales simultaneously, which we explore below.

Producing reliable, reproducible technology

According to the authors, the micro scale is where matter is engineered. At the micro scale, the venture often begins with a material, process, cell line, photon, or qubit that must be manufactured with extraordinary precision. Advanced semiconductors require vibration-isolated cleanrooms and ultra-pure process chemistry, tightly controlled biological processes and material substrates that must be reliable before anything is built on top of them. 

The common thread is that none of these can be de-risked cheaply, and the capital cost of building specialized facilities is often prohibitive for early-stage companies, which is precisely the constraint that our partner IXCAMPUS addresses by providing ISO 8 cleanrooms and shared equipment that allow hardware companies to compress the prototype-to-pilot timeline without capitalising the facility themselves. 

In biotech, the equivalent bottleneck is access to biofoundries and automation platforms capable of turning promising biological discoveries into reproducible and scalable processes which the HEC BioTech Stream is addressing through partnerships with specialized facilities across the European ecosystem. 

The meso scale is where systems are integrated. At this scale, the question changes and the technology needs to work not in isolation but as part of a system (often someone else’s). Battery cells must be packaged, managed thermally, and integrated with power electronics before they can go into a truck depot. A drone is not only a motor and a frame; it is a sensor suite, a flight controller, an edge AI system, and a communications link, integrated in a package that must survive thousands of flight hours under conditions a laboratory cannot replicate. A robotics venture must solve not just actuation and perception but the workflow integration that makes a robot useful inside a real factory. 

This is where partnerships become vital for de-risking technology and growth. CURA Climate’s team have got this point precisely: the team is developing an electrolyte platform to split limestone before the kiln to reduce CO2 emissions in cement production. They are not merely questioning whether the electrochemistry works in a reactor but how  it can be retrofitted into existing plants without requiring operators to replace existing capex intensive infrastructure, or accept production downtime that the economics cannot absorb. 

Nebu offers a digital version of the same meso-scale challenge. The founders began from a simple but serious pain point: startups trying to scale on cloud infrastructures are undermined by misconfiguration, technical debt, security gaps, and fragile operational foundations that turn what looks like a DevOps problem into a business problem. Nebu’s relevance here is that it treats cloud infrastructure as a system to be coordinated, supervised, and trusted, not simply as a technical layer hidden behind the product. 

This is why the ventures that navigate this scale most effectively are those that have identified early on industrial partners not as customers but as co-developers, to bring the operational knowledge and regulatory relationships that no venture would gain quickly enough to enter the market before they run low on funds. The connection to SMEs and centres of innovation that assemble those together, like the Heilbronn Campus matters here precisely because these family-owned industrial groups have solved system integration problems at scale, across decades, in exactly the sectors: automotive, industrial machinery, precision engineering where deep tech ventures need to prove their technology works in the real world.

Transforming infrastructure and institutions

It is at the macro scale that civilizational bets are made. At the top of the stack are the infrastructure systems whose replacement defines the energy and industrial trajectory of the coming decades.

Fusion energy is a useful example; if it reaches commercial viability, it will not simply plug into the existing grid but it will require a complete rethinking of baseload power architecture. Renaissance Fusion is developing high-temperature superconducting magnets for compact reactors, a bet that the path to fusion runs through better materials science, not only bigger machines. 

Space infrastructure faces a similar challenge: the falling cost of orbital access is creating the possibility of satellite constellations that provide the continuous, global monitoring layer that Earth-based sensors cannot obtain. As more companies and governments move infrastructure in space, keeping the orbital environment usable and trusted becomes critical. Kall Morris Inc is using robotics, AI and machine vision to capture objects in microgravity addressing an increasing concern: debris in space. Thanks to the collaboration with the ISS (International Space Station), the venture can validate its technology before customers and insurers trust a new technology coming into space.

The macro scale is where the timelines are longest, the capital requirements are highest, and the people you need are no longer scientists but project financers, regulatory scientists and diplomats. It is also increasingly where Europe’s sovereign industrial strategy is being contested.

Connecting areas of expertise to achieve successful scaling

The basic problem with human civilisation is that it already works, and that is precisely what makes significant transformation so difficult: it requires working inside, and often against, systems that already have infrastructure, language, incentives, and routines optimised for the world as it is rather than the world being built.

Increasing the scale of a project does not make it less of a specialist job; ambitious projects do not become less scientifically demanding as they grow, nor do they trade depth for complexity, but rather multiply depth by complexity, so that the same venture that requires nanofabrication precision at the micro scale also requires systems integration expertise at the meso scale and institutional negotiation capacity at the macro scale, with no substitution between them. 

Each domain community is optimised for its own area of expertise, which is why jargon typically exists, it makes things faster, as long as you stay within the community, but the undesired effect is that scientists, engineers, industrial operators, regulators, and project financiers are speaking mutually unintelligible languages while trying to build the same structure, and that coordination failure is the predictable result even when every party wants the same outcome.  The solution to this Tower of Babel problem has always been the same: skilled interpreters, people capable of understanding multiple fields in sufficient depth to bridge them and integrate them without losing the rigour that each demands. 

This is the future - and the bottleneck - of Deep Tech: enough people capable of understanding multiple fields in depth, to bridge them, and integrate them. Each institution, university, or corporation should have one goal: to move from a monolingual to a multilingual organization. 

Authors: Livia Kalossaka, Aymeric Penven, Alex Charbonné, Abdelalim El Hamichi, Patrick Kappel, Maryam Khademian