Why Deeptech Needs Patient Capital

The venture capital industry built its rhythms around software. Twelve-month sales cycles, eighteen-month runways, two-year paths to product-market fit. That cadence works well for SaaS and consumer apps, where the core risk is distribution and adoption, not physics or compute constraints. But deeptech doesn't fit neatly into that template, and trying to force it often destroys the very things that make it valuable.

Deeptech companies — those building at the intersection of fundamental science and commercial application — require a different kind of patience. Not the passive patience of waiting for the market to come around, but active patience: the willingness to stay engaged through multiple hardware iterations, regulatory approval processes, compute-limited training runs, and the slow, grinding work of turning a research insight into a deployable system. This requires capital that doesn't panic when a quarterly update shows technical setbacks rather than growth metrics.

The pressure to show rapid traction in early-stage deeptech is one of the most destructive forces in the ecosystem. We have watched founders pivot away from genuinely novel approaches because their investors started asking for revenue eighteen months in — long before the technology was ready for commercialisation. The result is a generation of "deeptech" companies that look like deeptech in their pitch decks but behave like SaaS companies in their roadmaps. The hard science gets shelved. The differentiation evaporates. The company competes on execution rather than defensibility.

Patient capital doesn't mean undisciplined capital. It means capital that applies pressure in the right places at the right times. Early on, that pressure should be scientific and engineering rigour. Can the core hypothesis be proven in a controlled environment? Can it be reproduced? Is the technical moat as deep as the founders believe? These questions require time and honest dialogue, not growth metrics. Later, as the technology matures, the capital should shift toward helping the company find its first commercial customers, structure pilot agreements, and build a revenue model that reflects the true value of what they've built.

At Nexum, our investment framework explicitly accounts for this. We don't apply uniform milestone timelines across our portfolio. We work with each founding team to define what success looks like at twelve, twenty-four, and thirty-six months — and those definitions are specific to the technology and the market, not borrowed from a SaaS playbook. For a computer vision company building industrial inspection systems, early success might mean a single pilot customer and a proven false-positive rate below 0.1%. For an AI infrastructure company, it might mean a published benchmark result and three design partnerships. Neither of these looks like conventional traction, but both signal that the company is on the right path.

The deeptech category is about to see its most important decade. The convergence of falling compute costs, maturing foundation models, and vast stores of industrial and scientific data is creating conditions in which research-grade capabilities can finally be deployed at scale. The companies that will define this era are being founded right now. They need investors who understand that the timeline for building something genuinely new is different from the timeline for shipping a feature — and who are prepared to stay the course.