The standard venture playbook starts with demand. Build the market slide first. Show the TAM. Prove someone will pay. That logic governs software, where shipping a working product is relatively straightforward and the real uncertainty is whether anyone cares. Deep tech inverts the model entirely. The demand is often obvious — governments want autonomous systems, grids need new energy sources, compute infrastructure requires faster chips. What is uncertain is whether the thing can actually be built. That single inversion changes everything about how to evaluate an investment: what risk looks like, when it retires, and what it means to survive it.
In our view, understanding this distinction is not academic. It changes which diligence questions matter and which ones are largely irrelevant.
Market Risk vs. Technical Risk: The Fundamental Divide
The risks facing deep tech companies are structurally different from those facing software investments. Market and customer risks dominate in software — is the market large enough? Does the product have product-market fit? — whereas technical and financing risks dominate in deep tech.
If a deep tech company actually manages to build a commercial quantum computer, it is a no-brainer that the market is more than large enough and that customers will clamor to use it. The same logic applies to reusable orbital launch vehicles, next-generation chips, and autonomous defense systems. Nobody needs a market slide for any of those. The U.S. Department of Defense, commercial satellite operators, and hyperscalers are already in line. The question is purely whether the engineering delivers.
Hardware often has clearer demand at the category level — of course people want cheaper energy or more capable defense systems — but much higher technical risk. The market may be obvious in the abstract, while the hard questions are whether the product can be made to work, manufactured economically, deployed safely, and sold through the right channels.
This is what the conventional "hardware is hard" framing misses. The difficulty is not the same kind of difficulty. It is harder in a different dimension — and that matters for how you price it.
Risk Retires at Milestones, Not at Revenue
The most important practical implication of the inverted risk model is that technical progress is measurable in a way that market progress rarely is. You can test whether an engine hits its thrust-to-weight specification. You can verify whether a chip achieves a target inference throughput. You can confirm whether a material holds its yield strength under operational stress conditions. Each positive result retires a discrete tranche of risk.
Deep tech companies face genuine technical risk — the science may not work, or may not work at commercially viable cost points. The mitigation is stage-gated investment based on technical milestones. This is how sophisticated capital should be deployed here: not in a single commitment against a vision document, but in tranches calibrated to demonstrated technical progress.
The pitch is not a vision; it is a de-risking roadmap. And from an investor's perspective, reading that roadmap accurately — knowing which milestones genuinely retire which risks — is the core competency the category demands.
Failure Looks Different
One underappreciated feature of the inverted model is that failure in deep tech tends to be early and loud, while failure in software tends to be late and quiet.
A software company can raise multiple rounds, grow a team to several hundred people, accumulate customer contracts, and still quietly fail to achieve the unit economics that justify its existence. The failure accumulates slowly, often invisible until a down round or a controlled wind-down years in. The market risk resolves gradually, and badly.
Software companies usually win by moving quickly through product and distribution uncertainty. Deep tech companies usually win by making a smaller number of harder, less reversible decisions correctly: assembling the right specialized team, choosing the right architecture, retiring the right technical risks, navigating regulation, financing the company with the right mix of capital, and turning their technical progress into a durable moat.
In deep tech, the physics either works or it does not. A launch vehicle that fails on the pad has failed obviously and early. A reactor design that cannot achieve net energy gain fails the moment the measurement is taken. That clarity is brutal in individual cases — but at the portfolio level, it means capital stops flowing to dead ends faster. The companies that clear their engineering milestones are genuinely worth backing.
The Team Slide Is the Market Slide
If technical milestones govern risk retirement in deep tech, then the credibility and completeness of the team executing against those milestones is the primary diligence variable. Not the serviceable addressable market breakdown. Not the customer acquisition model. The team.
Deep tech companies need different early employees. A software startup can hire generalist full-stack engineers early on and add specialists later. Deep tech companies usually need an extremely specific early team. You do not hire a generalist biologist or electrical engineer; you hire someone with direct, hard-earned experience in the specific therapeutic pathway, actuator design, RF system, or manufacturing process you are focused on.
Deep tech companies usually win by making a smaller number of harder, less reversible decisions correctly. For founders, this means deep tech rewards judgment more than motion.
The corollary for investors is that evaluating a deep tech company requires being able to assess whether a given technical architecture is sound and whether the team has actually solved comparable problems before — not just whether they have PhDs from the right institutions. The distinction matters enormously.
The Moat That Forms on the Other Side
The most structurally important consequence of the inverted risk model is what a successful outcome produces. Companies that clear genuine engineering milestones — that cross the valley of technical uncertainty — emerge into a competitive landscape that almost no competitor can reproduce quickly.
Reproducing a deep tech breakthrough requires equivalent scientific talent, specialized equipment, and often years of iterative research. This creates durable competitive moats that software startups rarely enjoy.
Software has low barriers to entry but weaker moats; hardware has high barriers to entry but more durable moats. The corollary of that asymmetry is significant: the companies that survive deep tech's front-loaded technical risk emerge into a position that is structurally harder to attack than any software category. Patents, process knowledge, manufacturing yield curves, and operational data accumulated over years of iteration compound into a position that a well-funded newcomer cannot simply buy their way into.
In software, defensibility usually comes from network effects, switching costs, or data moats. In deep tech, it often comes from patents, trade secrets, and freedom to operate. A company that does not own its core IP cleanly faces existential risk. And the offensive side matters just as much: that IP may be the only thing preventing a well-capitalized incumbent from replicating the work once the market is proven.
The capital markets are increasingly pricing this in. Global deep tech venture investment reached $48 billion in 2025, up from $18 billion in 2020. AI captured over a third of all global deep tech investment. Defense tech rode geopolitical tailwinds to its strongest year on record, posting 81% growth. The sectors attracting the most capital — autonomous systems, advanced compute, space infrastructure, next-generation energy — share the same underlying logic: clear demand, hard engineering, and deep moats for whoever gets there first.
Global military spending surged to $2.7 trillion in 2025, growing at more than twice the pace of the prior two decades. The trend is set to continue as governments seek technological sovereignty. Defense tech has become one of the most significant capital magnets in private markets. Roughly $19 billion flowed into aerospace and defense startups in 2025, nearly double the $10 billion raised in 2024.
Where This Lands
In our view, the inverted risk model is not a niche framework for defense and space specialists. It is the correct mental model for evaluating any company whose primary uncertainty is engineering rather than go-to-market. That set of companies is growing — and the capital flowing to them is increasingly coming from investors who have correctly identified that the diligence skill required is different, not just harder.
The implication is straightforward: if you are evaluating a deep tech opportunity using the market-first lens you would apply to a SaaS business, you are measuring the wrong thing. The market slide is not the key document. The technical roadmap is. And the team's ability to execute against it — milestone by milestone, in disciplines where errors are expensive and irreversible — is the question that determines the outcome.
What would change this view: evidence that demand in a given category is genuinely uncertain, not merely unproven at scale. That would restore conventional market-risk dynamics and require a different framework. For now, in the sectors where technical risk is the dominant variable, the correct investment question is simply this: can they build it?