European deep tech drew €15 billion in 2024 — nearly a third of all European VC. The valley of death between scientific proof and commercial product is still claiming companies at the same rate.

AI drug discovery that compresses early-stage research timelines. Generative AI and ML models trained on molecular structure data are genuinely compressing the hit-to-lead phase of drug discovery. For biotech investors, the question is whether the AI platform is genuinely differentiated, whether it is integrated into a proprietary biological assay pipeline, and whether the output quality is validated by downstream clinical evidence.

Computational platforms that replace expensive physical experimentation. In materials science, synthetic biology, and semiconductor research, AI simulation platforms that can screen candidate materials in silico before physical testing compress experimental cycles. Companies with proprietary experimental data to validate their simulations create a feedback loop that improves model accuracy with every physical experiment run.

Scientific AI that generates publishable, peer-reviewed results. A deep tech company whose AI platform has generated results validated through peer-reviewed publication has cleared a reproducibility bar that most competitors have not. This is a meaningful moat signal in a sector where scientific credibility directly affects partnership, licensing, and acquisition conversations.

TRL 4 presented as TRL 6. The valley of death runs from TRL 4 to TRL 7, and it is structurally underfunded in the European ecosystem — the EIC Transition programme had its core challenge tracks suspended in 2024. Companies that need to cross this valley will require capital that current investors may not be equipped to provide at the required check size.

IP transfer that looks clean but isn't. University IP assignment agreements are often written by technology transfer offices whose primary obligation is to the institution, not to the spin-out. Background IP licensed on a field-of-use basis, co-inventors who didn't sign the assignment, and government-funded research with publication obligations all create IP fragility that surfaces in exit due diligence.

Regulatory optimism that isn't grounded in pathway analysis. Deep tech companies with novel mechanisms — AI diagnostic systems, gene-editing therapies, novel biomaterials — face regulatory pathways that have no historical precedent. The EU AI Act's high-risk classification for medical AI, combined with MDR/IVDR obligations, creates a compliance intersection that regulatory bodies are still actively working to harmonise.