Episode 2: Predicting deep tech success | Predicting Startup Success podcast

Predicting the success of a startup is never easy. Predicting the success of a deep tech startup is harder still.

Built on advances in science and engineering, deep tech ventures often require years of development, substantial capital and specialised expertise before their commercial potential becomes clear. Traditional indicators of startup progress, such as customer traction or revenue growth, may be absent for long periods. Yet investors, universities and policymakers must still make decisions about which ventures to support and which opportunities to pursue.

In conversation with Isabell Welpe in Episode 2 of the Predicting Startup Success podcast, Ramana Nanda, Professor of Entrepreneurial Finance and Associate Dean for Enterprise at Imperial Business School, argues that predicting deep tech success is not about finding a single decisive signal. Instead, it is about understanding uncertainty, how it evolves, and how different actors work to reduce it over time.

Deep tech is not one prediction problem

When evaluating startups, it is tempting to think primarily about market uncertainty.

Will customers want the product? Will the company find a scalable business model? Will growth materialise?

Deep tech ventures face those questions too, but they carry an additional layer of complexity. Before anyone can determine whether a market exists, they often need to establish whether the technology itself can deliver what it promises.

As Nanda explains, technical risk and market risk frequently interact. A technology may have multiple potential applications, each requiring different technical milestones, different customers and different paths to commercialisation. A change in direction is therefore not simply a strategic decision. It can mean revisiting fundamental assumptions about what needs to be proven and how.

This helps explain why the journey can be so difficult to predict. Success depends not on overcoming a single hurdle, but on navigating several uncertainties at once.

When laboratory success is not enough

One of the recurring themes in the conversation is the distinction between scientific achievement and commercial viability.

Academic research is often designed to establish novelty. Investors, by contrast, are trying to understand whether a technology can create value at scale.

As Nanda notes, “the key element that’s being optimised is novelty.”

The challenge is that the conditions that demonstrate an impressive result in a laboratory may not reveal whether a technology will succeed in the real world. A process that works at small scale may prove difficult to manufacture economically. A material that performs well under controlled conditions may behave differently in industrial environments. A proof of concept may answer one question while leaving others unresolved.

For this reason, the most important experiments are not always the most impressive ones.

What matters is whether they provide meaningful evidence about the questions that will determine future success. The goal is not simply to demonstrate that a technology works, but to understand whether it can work under the conditions required for commercial adoption.

Financing is a way of learning

Much of the startup world treats financing as a mechanism for providing resources.

Nanda offers a different perspective.

Rather than viewing venture capital solely as a source of funding, he argues that it should be understood as a process of learning. Each financing round gives founders and investors an opportunity to gather information, reduce uncertainty and reassess what comes next.

“If you think about multi-stage financing as a sequence of experiments,” he explains, “the key question becomes how do you generate more information from these experiments for less money?”

Viewed this way, milestones become more than markers of progress. They are pieces of evidence.

The most effective founders are often not those who simply hit targets, but those who understand what future investors will need to see and design experiments that answer those questions. Likewise, successful investors are not merely allocating capital. They are helping ventures discover which assumptions hold true and which need to change.

Prediction, in this context, becomes a process of reducing uncertainty rather than eliminating it.

Why innovation is a relay race

Towards the end of the discussion, the focus shifts from individual startups to the broader ecosystem that supports them.

Universities, technology transfer offices, seed investors, venture capital firms and later-stage investors all influence a venture’s trajectory. Yet they often operate according to different incentives and priorities.

For Nanda one of the biggest challenges is that organisations frequently concentrate on the stage directly in front of them rather than the stages that follow. A startup may secure early funding, but still struggle later if it has not generated the evidence required by growth-stage investors. Similarly, a university may help develop promising research without fully considering the financing pathway needed to bring it to market.

To explain this, he uses a simple analogy:

“Each of us need to run our relay race where we need to run towards where the next person is standing.”

The image captures a broader point about commercialisation. Progress depends not only on how well an individual contributes, but on how effectively they prepare the venture for the next stage of development.

Rethinking prediction

By the end of the conversation, deep tech prediction looks rather different from the search for a perfect signal.

The question is not simply whether a technology appears promising today. It is whether founders, investors and institutions are generating the evidence needed to understand what that technology might become tomorrow.

For anyone trying to evaluate deep tech ventures, that may be the most important insight of all. Success is rarely revealed in a single breakthrough moment. More often, it emerges through a series of decisions, experiments and commitments that gradually bring an uncertain future into clearer focus.