A Framework for Evaluating Deep Tech Startups

July 2020 • FailUp Capital Investment Team

The frameworks that venture capitalists use to evaluate startups were developed primarily in the context of software businesses. They prioritize metrics that matter for software: user growth, engagement, revenue run rate, net revenue retention, payback periods. They reward businesses that can scale without proportional increases in cost structure — the marginal cost of serving the next customer approaches zero. And they operate on timelines calibrated to software: meaningful revenue in eighteen months, market leadership in five years, exit in seven to ten.

Applying these frameworks to deep tech startups — companies built on scientific discoveries in materials, biology, physics, or chemistry, and requiring hardware development, manufacturing scale-up, and rigorous customer qualification processes — produces systematically wrong answers. The result has been an underinvestment in deep tech relative to its economic importance, a misalignment of expectations between investors and founders, and a graveyard of deep tech companies that ran out of money at the worst possible moment: just before the science was ready to translate into commercial value.

At FailUp Capital, we have built our investment practice around a framework specifically designed for deep tech. This essay lays out that framework — not as a rigid checklist, but as a set of principles for thinking clearly about science-driven companies.

Principle 1: De-Risk the Science First

In software, the science is never the risk. Writing code works; the question is whether anyone will use what you build. In deep tech, the science is often the primary risk. Does this battery chemistry actually achieve the theoretical energy density? Can this material be synthesized reliably enough to build a product? Does this process produce the yield necessary for commercial economics?

The first question we ask about any deep tech company is: what is the current state of scientific risk? If the company is still trying to demonstrate the fundamental scientific principle, it is too early for FailUp Capital — we are Seed stage investors, but we need to see scientific proof of concept before we invest. We are looking for companies where the core scientific question has been answered — in the laboratory, in peer-reviewed work, or in a credible series of controlled experiments — and the primary challenge is now engineering: how to take a working scientific demonstration and turn it into a product.

This distinction between scientific risk and engineering risk matters enormously for timeline and capital planning. Engineering challenges — even very hard ones — tend to be reducible by throwing talent and capital at them. Scientific challenges are different: there is no amount of capital that can force a materials discovery to happen on schedule. If a company is fundamentally still searching for its core scientific breakthrough, the investment timeline is genuinely unpredictable.

Principle 2: Evaluate the Team for Both Technical Depth and Commercial Breadth

Deep tech founders face a unique team-building challenge. The scientific and engineering expertise required to build a breakthrough hard tech company is rare, specialized, and often found in people who have spent ten or more years in academic or government research environments. These individuals have extraordinary technical depth but may have limited experience in sales, customer development, manufacturing operations, supply chain management, or financial planning.

FailUp Capital looks for founding teams that have both dimensions covered — either in the founders themselves or in early hires. The best deep tech companies we have seen have a technical co-founder who owns the science and a complementary co-founder who owns the commercial development. Neither can be a passenger; the technical founder needs to understand the market deeply enough to know which scientific choices create commercial value, and the commercial founder needs to understand the technology deeply enough to identify the right customers and the right applications.

We are also evaluating intellectual honesty. Deep tech founders spend years in academic environments where skepticism and self-criticism are virtues. The best founders we meet are those who can tell us clearly what they do not yet know, what could go wrong, and what assumptions their plan depends on. Founders who cannot identify the central technical risks in their own company are not yet ready to manage them.

Principle 3: Map the Customer and the Qualification Path

Deep tech products rarely sell through consumer channels or self-service SaaS funnels. They sell into industrial, defense, healthcare, or infrastructure markets where procurement processes are formal, qualification testing takes time, and the buying committee includes engineers, procurement professionals, and finance officers. Understanding this go-to-market reality is essential for capital planning.

When we evaluate a deep tech company's commercial strategy, we are looking for a realistic answer to a specific question: who will pay us money, how much, and when? The answer should include named customer targets (or named customers in active discussions), a realistic timeline for qualification testing, and an understanding of the specific performance thresholds the customer requires. Abstract market size estimates and ideal customer profiles are insufficient; we want to see evidence that specific real humans at specific real companies have expressed willingness to pay for this product at these performance levels.

We also pay close attention to the qualification path. In aerospace, qualifying a new material or component can take three to five years and cost millions of dollars in testing. In medical devices, regulatory clearance is a multi-year process with uncertain timeline. In industrial chemicals, customer testing and qualification processes are standardized but slow. Understanding these timelines — and having capital to survive them — is non-negotiable for any deep tech company in these sectors.

Principle 4: Understand the Manufacturing Scale-Up Challenge

The history of deep tech is full of companies that proved their technology worked in the laboratory but could not make the economics work at manufacturing scale. The transition from laboratory-scale production to commercial-scale manufacturing is often as technically challenging as the original scientific breakthrough — and it is frequently underestimated in both time and capital requirements.

We evaluate manufacturing scale-up risk explicitly in our investment process. What process produces the technology at laboratory scale? What are the known failure modes when that process is scaled? Has the team benchmarked the likely yield, cycle time, and capital cost of commercial-scale production? Are there precedents for this type of manufacturing scale-up, or is it genuinely uncharted territory? These questions do not have easy answers in the early stages of a company, but a founding team that is thinking seriously about them — rather than deferring the question to a later date — is a much better bet than one that has not engaged with the manufacturing challenge at all.

Principle 5: Model for Realistic Timelines

One of the most persistent errors in deep tech investment is optimistic timeline modeling. A company that realistically needs seven years and $80 million to reach commercial scale will struggle to survive if it is capitalized for four years and $20 million. When the timeline inevitably slips — as it almost always does in hardware and science-driven businesses — the company enters a fundraising environment where it has less to show than it promised, making the next round more difficult and more dilutive.

At FailUp Capital, we push back hard on optimistic timelines in investment plans. We have developed internal benchmarks for how long different types of deep tech development milestones typically take, based on our experience and our review of comparable companies. We use these benchmarks to reality-test founder projections and to size our investment to support a company through at least two development milestones — not just to the next fundraising date.

Key Takeaways

• Software venture frameworks systematically misapply to deep tech; the industry needs purpose-built evaluation criteria for science-driven companies.
• De-risking the core science before Seed investment is FailUp Capital's primary filter; engineering risk is manageable with capital, scientific risk is not.
• Deep tech founding teams need both technical depth and commercial breadth; intellectual honesty about known unknowns is a strong positive signal.
• The customer qualification path — including named targets, specific performance thresholds, and realistic timelines — is a non-negotiable element of commercial due diligence.
• Manufacturing scale-up is typically underestimated in both time and capital requirements; founders who have engaged seriously with this challenge are stronger bets.