The venture capital ecosystem is shifting. Digital moats are eroding, and generalist and SaaS VCs are turning their attention to complex science and physical engineering. By now, deep tech has firmly established itself as a core asset class. Globally, venture investment in the sector reached $48 billion in 2025, a massive leap from $18 billion in 2020.
Nowhere is this transition more visible than here in Europe, where the aggregate valuation of venture-backed deep tech enterprises has hit a record $690 billion. Driven by a fundamental inversion of margin dynamics between software and hardware, the UK alone has attracted $43.7 billion since 2019, ranking third globally. Across Continental Europe, deep tech investment reached $20.3 billion in 2025, accounting for an all-time high of 32% of total European VC, with France and Germany as leading the hubs.
Whether you are looking at London for novel AI, Munich for defense and space, or Zurich for advanced compute, evaluating these companies requires a completely different playbook.
Even with a background in biochemistry, five degrees across several other disciplines, a career spanning fintech and startups, and hands-on experience helping a grant-funded biotech data infrastructure company build a sustainable path to revenue, my first deep tech due diligence last year was not as I imagined. I got bogged down in technical nuances that were ultimately so unknown at that stage that I should have just acknowledged them as an inherent risk. I’ve learned a lot since then, so I wanted to share a few of those lessons here.
LLMs won’t cut it: First, identify the real bet
You could try using ChatGPT, Gemini, Claude, or Perplexity to draft your investment memo or handle the background research, but they are never quite good enough on their own. You cannot simply let an LLM run wild with a data room and expect a flawless DD report. Using these tools effectively requires substantial pre-work and a lot of iterative prompting.
To make them useful, you have to do the strategic heavy lifting first. When evaluating a deep tech startup, you must truly isolate the primary risk. What is the fundamental bet you are making? Is it a bet on the underlying science, the engineering scalability, or the market adoption? Your entire due diligence direction depends fully on this answer, and this clarity is exactly what must drive your work with the LLM.
Commercial traction: Look for desperation, not just plans
You are not looking for a polished business model or a multi-year plan. The single most important point of evidence for a deep tech company is proof that it is not just an elegant science project. You need to know that the market timing is right and that the founders understand who their customers are.
Better yet, they need to identify the single individual who makes the buying decision and find out if they are desperate for this specific solution right now.
I know what you might say. But this is a pre-seed or seed opportunity; they don’t have customers yet. As the computer scientist turned deep tech investor and founder of Silicon Roundabout Ventures, Francesco Perticarari pointed out:
On one end of the scale, you have Nuclear Fusion or Fault-Tolerant Quantum Computing: Nobody has proven, even in a lab, that these are 100% possible yet. It is a “reasoned guess.” If you build a net-energy positive reactor tomorrow, you have a queue of governments and data centres waiting to buy. The “hunger” for limitless, clean energy is immediate. The science is the risk, but the market is a certainty.
Expressions of interest and LOIs without binding contractual clauses simply are not good enough anymore, either for you or for the founder. If the founders are not finding the customer who is happy to sign an LOI with a commitment to buy after a certain milestone is cleared, they are just wasting everyone’s time. Founders must apply the same hustle to validating their market as they do to developing their technology.
De-risking the academic execution gap
According to the 2026 European Deep Tech Report by Walden Catalyst Ventures, Lakestar and Dealroom, research spinouts account for 33% of new European deep tech startups since 2015, particularly in photonics, quantum technologies, and advanced computing. The typical European deep tech founder has a highly academic background, often holding a master’s degree or PhD, with a median age of 35. However, only 24% have prior startup experience, while 21% come from corporate or consulting environments.
This lack of commercial operating experience naturally increases execution risk and stretches commercialisation timelines. To assess the technology, the team must demonstrate that they can actually make it work and defend it scientifically.
A useful mental exercise is to project forward to the Series A round. Look at what Series A investors in that specific vertical are demanding today. By understanding those future expectations, you can internalise the milestones your seed or pre-seed startup must acquire within the timeframe of the current raise. Can they achieve these proof points with less money? If the answer is a convincing yes, the funding amount is sufficient. Your goal as an investor is to construct and understand the technical de-risking map for the company and match that to the founders’ abilities.
The cap table: Non-dilutive funding is a necessity
In Europe and the UK, grant funding is not just a bonus; it is an industry standard. With the European Innovation Council (EIC) deploying €1.4+ billion across various programmes and the UK allocating £7 billion to support innovative company growth, non-dilutive capital is a highly positive signal. It allows founders to achieve critical milestones without prematurely giving away equity.
In fact, if a deep tech company based in the UK or Europe has zero grant funding, it might send a negative signal. It forces the question: why not?
Looking back at that first due diligence, I thought that my biggest challenge was understanding the science. However, understanding the science is not the whole job. The job is figuring out which bet you’re actually making, then building the map that tells you whether these founders can clear it with the money they’re raising. The technical rabbit holes were a way of avoiding that harder question, not answering it.
That’s really what all four points come down to. Isolate the real risk. Hunt for the customer who is desperate, not just curious. Reverse-engineer the Series A milestones. Read the cap table for what the grant funders already believe. None of it requires you to be the smartest scientist in the room (you won’t be). It requires you to be honest about what you’re betting on and disciplined about how you’d know if you were wrong. We can certainly spend much more time exploring the nuances of IP litigation, lab space access, and supply chain constraints, but this hopefully provides a solid starting framework.
The capital is moving this way whether or not we’re ready for it. The investors who do well won’t be the ones who focus just on the science. They’ll be the ones who always ask better questions.
Only 15% of seed funding goes to women-led deep tech start-ups. Discover the research behind the numbers.
Thinking about launching a VC fund? Explore a practical playbook covering fundraising, LPs, fund structure, and portfolio strategy.
