thoughts on raising a deeptech preseed
from first-time founder principles
Deep-tech companies are in the business of building radically new capabilities for humanity1. Using metrics for success like revenue, traction, and customers early-on usually doesn’t make sense. This is primarily because a new capability, by definition, is something that hasn’t been priced-in by the market, existing players and numbers are not yet equipped to engage or evaluate its true potential.
This doesn’t mean that market, and GTM doesn’t matter, in some ways, it matters even more, especially GTM because deeptech R&D is expensive. However, it does mean that the better questions are usually directed towards
i) how painful and widespread the problem is,
ii) by how much (ideally at least 10x) and on how many dimensions (ideally at least two) does the new capability solve for the problem,
iii) what are the unfair advantages that the company possesses to make this new capability a reality and continue to dominate its blue-ocean, like intersectional adeptness in the founding team, integral know-how with compounding returns, or piercing unintuitive insight.
The first gives you a sense of the market, the second the likelihood of adoption (and margins), and the third the winning-ness.
All three of the above could be home-runs but if you do not address GTM, it might not mean much, because you’ll run out of money before you win. But deeptech GTM differs from say SaaS GTM, in that the goal isn’t to balloon outwards the SOM SAM TAM by driving in a wedge and then expand outwards, but rather, to simply make some money to foot bills and avoid dilution (and failure), increase runway between milestones and more importantly, be in the arena and win “battle-tested” points. Battle-tested can mean different things depending on the domain - clinical data validation, flight time, manufacturing capacity - but these points are generally critical to position you to take much bigger swings, including big contracts or grants. Crucially, this means GTM for deeptech is less concerned about the initial customer or contract size, though a juicy contract doesn’t hurt, but more on the ability to amass high-leverage battle-tested points.
Understanding competition is important, though again with a different lens. They’re two broad kinds in deeptech - ones working on unlocking the same capability, and ones unlocking a different capability with overlapping applications. We can think of the first kind as IP competition and the second as business competition (h/t Spencer Schneier). With IP competition, it’s important to understand (and convince) why your specific tech roadmap and understanding of the science is the best way to overcome challenges others have and may continue to face. With business competition, you have to hone in on why your particular capability has an x factor(s) in the right axis to capture the market, over another capability. These are two very different cases that shouldn’t be conflated.
Almost all variables a founder cares about - valuation, timeline, and odds of getting a check - are tied to risk. Or perhaps more accurately, the perception of risk. The more risk that appears in need of underwriting, the lower the company is worth, the slower things move, and the less likely you get funded. In theory, deeptech investors should be more worried about product risk - will the founding team actually crack the code of this unsolved problem? In practice, it may not always be apparent why cracking the code is important enough to make a lot of money.
A de-risking map seems important and many founders apparently fail to convey it (h/t Manoj Ramaiah saying it takes 3 meetings to fish it out). Khosla has a list of questions of what a good map lays out. While this seems clearly vital for seed, it’s not as clear to me how crucial or even possible this is for preseed, where PMF is yet to be achieved and risk can be pivoted away. It might be because I haven’t thought about this hard enough.
Even though it sounds counterintuitive, “Romantic” founders may win in deeptech comparatively more often than “Pedigree” founders (h/t Edward Lando), compared to other sectors. While they’re definitely domains where deep and sustained expertise such as that gained from a PhD is a necessity, such as nanomaterials, and intense subject fluency is almost certainly a prerequisite in every domain, too much founder pedigree could backfire by artificially constraining the possibility space and curtailing the degree of ambition in the narrative. The Romantic founder is willing to commit to a gut hunch without yet having a very evidence-backed reason to do so, which is vital to bet and win on trends that are still emergent.
Finally, what even is deeptech? Matthew Mandel describes three kinds of plays that are generally clubbed together - hardware, infra, and frontier (h/t to Divyansh Saksena). I think frontier deeptech captures the definition I’ve been speaking to in this piece and “building radically new capabilities for humanity” seems to capture this well.
or dogs (and eventually dolphins)