From Description to Decision in Deep Tech

From Description to Decision in Deep Tech

Why defining deep tech is not enough — and what governance actually requires

Witte, M. K. (2026), DOI 10.5281/zenodo.19959238

Everyone is talking about deep tech. But very few institutions can actually decide what to do with it.

Billions in public funding, mandate renewals, KPI structures, and investment decisions depend on classifying something as "deep tech." Yet the frameworks used to do this — even the most rigorous ones — consistently stop at description. They tell you what deep tech tends to look like. They do not tell you when something qualifies, how to handle borderline cases, or what the classification requires you to do differently.

This paper names that gap precisely — and shows why it matters far beyond academic semantics.

What this paper argues

Recent attribute-based frameworks have improved the vocabulary of deep tech innovation. They consolidate recurring characteristics: scientific origin, hardware intensity, long development timelines, high capital intensity, platform character, multilayered uncertainty. That is useful work.

But it is not enough.

A framework that cannot decide boundary cases is not yet operational.

A taxonomy that cannot model change over time is not yet a theory of behaviour.

A definition that does not change what an institution does next has not yet become a governance instrument.

This paper evaluates the current state of deep tech frameworks against four structural requirements: threshold specification, boundary resolution, operationalization, and dynamic validity. The result is consistent: the field is descriptively rich and operationally thin.

What you get in this paper

The Decision-Capability Test (DCT) — a public, five-criterion evaluation instrument introduced in this paper. It assesses any deep tech framework against threshold, boundary, operationalization, dynamic, and governance criteria. Scored on a 0–2 scale per criterion. Total maximum: 10. Most current frameworks score between 2 and 4.

A Framework Evaluation Matrix — scoring of BCG/Hello Tomorrow, Romasanta et al., Cabanes, Kortsch et al., and TRL/MRL systems. Side-by-side. Interpretive. Transparent about what each framework was designed to do and where it stops.

Six boundary-case stress tests — AI-enabled robotics, quantum software, university spin-offs with incremental tools, advanced materials services, corporate lab innovation, industrial AI optimization. Each case shows exactly how current frameworks describe ambiguity without resolving it.

A Decision-Maker Utility Table — what attribute frameworks tell investors, accelerators, TTOs, public funders, corporate innovation units, and oversight bodies. And what they still cannot decide after reading the framework.

A quantified asymmetry — in the most extensive current consolidation framework: 12 defined attributes, 3 layers, 4 adjacent concepts, 25+ future research questions. Hard quantitative thresholds: 0. Governance decision rules: 0. The asymmetry is structural, not a failure of execution.

A concrete research agenda — threshold research, boundary research, operationalization research, dynamic research, governance research. Five directions for a field that needs to move from description to decision.

Who this is for

- Programme leaders and managers running deep tech incubators, accelerators, TTOs, and corporate innovation units who need to explain why their KPI structure reflects the actual nature of what they support

- Public funders and oversight bodies who are expected to demonstrate long-term outcomes but work with definitions that do not support measurement

- Innovation policymakers designing instruments for deep tech support who need to understand why adding capital without fixing measurement architecture reproduces the same distortions at larger scale

- Researchers working in innovation management, science commercialization, or technology policy who want a precise diagnostic of where the definitional literature currently stands and where it falls short

- Investors and venture builders who want a clearer conceptual foundation for distinguishing deep tech development logic from software or digital innovation logic

What it does not do

This paper does not propose another definition of deep tech. It does not offer a consulting framework, a playbook, or a step-by-step guide. It is a conceptual position paper — rigorous, evidence-based, and direct.

The Decision-Capability Test is offered as a public, non-proprietary evaluation lens. Academic citation and non-commercial use are permitted with attribution. Commercial use, derivative tool development, and institutional deployment require prior written permission.

The author's broader work

The author has developed a concrete decision system — the 4×4-TETRA Deep Tech Matrix™ and its institutional governance layer — that operates at the model and governance layer this paper identifies as missing. That system has been independently validated as original R&D under OECD Frascati Manual criteria by the German Federal Ministry of Research, Technology and Space (BMFTR). Its operational methodology is proprietary. This paper is the public conceptual layer that explains why such a system is needed.

Publication details

**Author:** Maria Ksenia Witte, Arise Innovations®, Berlin

**Published:** 01 May 2026

**License:** CC BY-NC-ND 4.0

**Format:** PDF, 18 pages + appendices

**Language:** English

Deep tech has become more describable. This paper explains why that is not the same as governable — and what the field needs next.