Deep Tech Rising
Smart capital is shifting from software to frontier technologies with urgent buyers
The venture capital industry stands at an inflection point. After two decades of pouring capital into software startups, investors are discovering that the easy wins have been claimed. Software’s promise of infinite scalability and minimal marginal costs has collided with brutal realities: low barriers to entry, fierce competition, and the emergence of AI tools that commoditise code itself. Meanwhile, AI, hailed as the next frontier, demands capital at a scale that favours megafunds and tech giants, leaving most investors facing uncertain returns in what many suspect is a bubble.
This shift has prompted a fundamental reallocation of capital towards deep tech—ventures that combine cutting-edge science with tangible, physical outputs. These companies build batteries that transform transport, materials that reshape manufacturing, and systems that modernise defence. Unlike software, which can be copied almost overnight, deep tech creates defensible moats through patents, process knowledge, and capital-intensive infrastructure. And unlike AI, where incumbents dominate, deep tech offers genuine opportunities for startups to create entirely new industries.
Yet deep tech challenges conventional venture wisdom. It demands founders with deep technical expertise, not just customer insight. It requires long-term vision paired with relentless pressure for early milestones—prototypes, operational metrics, and customer adoption that validate progress. It needs investors who understand both equity and debt, science and manufacturing, innovation and industrial scale, startups and big corporations. Most importantly, it shifts the primary risk from market acceptance to technological achievement—if you can build it, buyers are already waiting.
This edition examines why tech investors are abandoning traditional software for deep tech, how this transition reshapes risk assessment and capital deployment, and what it means for the future of innovation. Through ten interconnected arguments, it reveals that deep tech is not simply another investment category but a return to venture capital’s original purpose: funding breakthrough technologies that transform industries and create lasting competitive advantage.
For those willing to master its distinct challenges, deep tech offers what software no longer can—the opportunity to build companies that are both economically defensible and strategically vital. A reading list follows for those seeking to explore these themes further.
1/ Tech investors are abandoning traditional software and focusing on deep tech
For them, the era of deploying entire funds into SaaS startups is over because AI has upended the market. Investing in AI startups is costly—you are up against US megafunds and big tech incumbents, while everyone warns it is all a bubble anyway. As a result, like those who backed Hopin or Getir in 2020 and 2021, investors in AI today may never see their money again. As Jerry Neumann wrote in his landmark article, AI will not make you rich:
Investing in the maturity phase is… difficult… nothing much happens at all. The uncertainty about what will work and how customers and society will react is almost gone. Things are predictable, and everyone acts predictably.
All in all, traditional software is waning, and AI is high-risk (high beta). And so deep tech it is.
The label is vague—some call it hard tech, frontier tech, industrial tech, or production tech—but its meaning is clear: (i) it is scientifically cutting-edge, achievable only through rigorous lab research, and (ii) it is tangible, producing outputs that require factories, clean rooms, wind tunnels, or other physical assets.
There is a logical explanation for deep tech’s rise. It touches domains that software has barely reached because software needed digestible entry points, like electric cars or smartphones. Before software can truly transform a car, the car itself must become electric. Otherwise, software is just a navigation tool measuring average consumption. Electric cars turn software into an operating system that controls everything—power, propulsion, and embedded features. Thus deep tech, for example in batteries, is technology’s gateway to dominating the multi-trillion-dollar car industry.
There is more. To paraphrase David Galbraith, software may have eaten the world, but now hardware is eating software back. Unit economics matter in a competitive environment. If your product is more than software—an entire car operated by software—then software is important, but it is not where your company gains an edge. Software has near-zero marginal cost, and open-source components make advantage hard to sustain. As David observes, the rational choice is to price software at zero and compete on reliability and yield on the hardware side, counting on superior manufacturing that operates at a larger scale. That’s exactly what the Chinese do 🇨🇳.
Conversely, imagine an EV company cutting hardware costs while charging more for superior software. The result is a poor-quality car that constantly needs repair, paired with software that can never prove its worth because the underlying hardware is unreliable.
All in all, deep tech matters more than ever because it combines scientific breakthroughs with tangible execution, creating a defensible advantage where software alone cannot anymore. It is where true, long-term value in technology lies. This is why investors are into deep tech these days.
2/ Founders and investors face tougher challenges when building deep tech companies
From a founder’s perspective, deep tech is not for the faint of heart, because unlike software, it is harder to start from the customer’s needs:
In software, you can be fresh out of university or a coding bootcamp, or even know little about software itself. All you need is the ability to spot a user’s need and build a tool to meet it (“Make something people want”)—by coding the MVP yourself, partnering with a competent technical co-founder, or using tools like Lovable or Cursor. If the business takes off, you can always hire stronger talent later to serve users at scale.
In deep tech, it is different. You might imagine the perfect product, but bridging the gap between concept and a sustainable, working prototype is far harder without deep industry knowledge and years of lab or factory experience. Building Airbnb was about discovering what people wanted—turns out, borrowing someone else’s sofa was it. Building a deep tech product, by contrast, demands far more technical skill and discipline.
For investors, the picture looks even worse. Software used to be reassuring for VC firms because the path from inception to exit, however challenging, was clearly laid out. Deep tech, by contrast, is seen as too capital intensive, highly dilutive for early investors and founders, slow to exit, and incompatible with Silicon Valley’s standard model of funding through seed, Series A, B, and so on.
Yet the data tells a more nuanced story. Leo Polovets’s recent analysis of PitchBook data for US and Canadian companies shows that deep tech firms are sometimes less capital intensive than purely software-based ones. To see why, we need to remember that every company sits somewhere on the scalability–defensibility spectrum—and there is no single “optimal” position when it comes to capital intensity.
A company must be scalable because capitalism rewards increasing returns, allowing firms to escape competitive pressure. But scalability alone is not enough if the business is not defensible. Consider airlines: they benefit from network effects, yet margins remain thin because competition persists at any scale. The same applies to oil: rising prices make drilling in expensive locations viable, which increases supply and ultimately stabilises the market.
Software faces a comparable dynamic. If a software company raises prices, competitors rush in because barriers to entry are so low in that space. As Jeff Bezos famously put it, “Your margin is my opportunity.” Low barriers to entry mean rivals will appear quickly, forcing companies to raise more capital to maintain an edge—sometimes exceeding the capital intensity of deep tech firms.
It takes exceptional characteristics, such as powerful network effects across multiple sides of the business (as Google enjoyed for years), to keep competitors at bay. For most software companies, though, life is not much easier than for deep tech ones. Success is far from guaranteed because they must raise significant capital, exits are uncertain, and competition remains fierce all the way up.
3/ Markets reward companies that combine defensibility with capital efficiency
Mostly, it’s because most competitors crash into the funding wall. If Company A has raised a lot but hasn’t tightened the bolts or improved its unit economics, it remains dependent on new capital to subsidise demand. When that capital dries up—whether due to a shift in the macro environment or investors simply losing patience—the prize doesn’t go to the company that raised the most. It goes to Company B, which resisted overreliance on funding and achieved the highest capital efficiency.
As Kai-Fu Lee wrote in AI Superpowers:
From the outside, venture-funded battles for market share look to be determined solely by who can raise the most capital and thus outlast their opponents. That’s half-true: while the amount of money raised is important, so is the burn rate and the “stickiness” of the customers bought through subsidies. Startups locked in these battles are almost never profitable at the time, but the company that can drive its losses-per-customer-served to the bare minimum can outlast better-funded competitors. Once the bloodshed is over and prices begin to rise, that same ruthless efficiency will be a major asset on the road to profitability...
Wang Xing [founder of Chinese delivery startup Meituan] embodied a philosophy of conquest tracing back to the fourteenth-century emperor Zhu Yuanshang, the leader of a rebel army who outlasted dozens of competing warlords to found the Ming Dynasty: “Build high walls, store up grain, and bide your time before claiming the throne.” For Wang Xing, venture funding was his grain, a superior product was his wall, and a billion-dollar market would be his throne.
More generally, defensibility acts as a buffer against endless fundraising. Sometimes it comes from true scale, which delivers better economics through efficiencies and allows companies to beat competition once and for all. In other cases, it stems from lock-in, access to key resources, superior branding, or other advantages. The list is long, but behind every successful company are two recipes: one for scaling the business, the other for making it defensible.
If you have scalability without defensibility, your company becomes a profitless capital sink and capital intensity soars. If, on the other hand, defensibility emerges early enough to reduce the need for additional funding, capital intensity stays under control, and investors can expect a return.
And this seems to be exactly what is happening with deep tech. The aforementioned Polovets analysis shows that deep tech companies, at least in North America, are often less capital intensive than less-defensible software companies. In other words, in software, markets can stay irrational longer than startups can stay solvent—but deep tech may be where rationality finally pays. And this dynamic—where capital efficiency and defensibility determine long-term survival—helps explain why deep tech may be less capital intensive than it seems.
But understanding capital needs is only part of the story. To evaluate deep tech properly, investors must also consider the full spectrum of risk: technological, market, financing, and management. That is where the profile of deep tech becomes truly distinctive.