Back to Atoms
For the past two decades, venture capital gravitated toward software — specifically, enterprise SaaS and consumer internet. The reasons were rational: software has high gross margins (70-90%), scales near-infinitely, requires relatively little upfront capital, and can achieve product-market fit quickly. Hardware, biotech, and energy — the “deep tech” sectors — were the opposite: capital-intensive, slow to develop, risky, and requiring specialized expertise that most VC generalists lacked.
That’s changing. Deep tech investment reached roughly $100 billion globally in 2024, according to BCG and Hello Tomorrow, up from about $40 billion in 2019. The share of VC dollars flowing to deep tech has roughly doubled. The drivers are a combination of genuine technological breakthroughs, massive government incentives (the IRA, CHIPS Act), and the realization that the highest-impact — and potentially highest-return — opportunities increasingly lie outside pure software.
Climate Tech: The IRA Effect
The US Inflation Reduction Act (2022) committed roughly $370 billion to climate and energy programs over a decade. The impact on climate tech investing has been dramatic. The IRA’s tax credits — for renewable energy, battery manufacturing, hydrogen, carbon capture, and advanced nuclear — provide revenue certainty that changes the risk calculus for investors.
Notable deals: Commonwealth Fusion Systems, an MIT spinout developing commercial fusion energy, raised over $2 billion (including from Bill Gates, George Soros, and Tiger Global) — one of the largest private climate tech investments ever. Form Energy raised $450 million for its iron-air batteries designed for multi-day grid storage, something lithium-ion can’t economically provide. Redwood Materials (JB Straubel’s battery recycling company) raised over $2 billion and is building a massive cathode materials campus in Nevada.
Climate tech exits are happening too. Occidental Petroleum acquired Carbon Engineering (direct air capture) for $1.1 billion in 2023. Nest acquired Maji Equity-backed Essentium (a comparison platform) — demonstrating that strategic acquirers are willing to pay for climate tech. The exit pipeline is thin relative to the investment pipeline — most climate tech companies are pre-revenue or early-revenue — but the thesis that climate tech can generate venture-scale returns is being validated in real transactions.
Hardware Renaissance
The “hardware is hard” mantra scared VCs away from physical products for years. But several trends have made hardware more venture-friendly: the maturation of contract manufacturing in Asia means prototypes can be productionized faster and cheaper; crowdfunding platforms (Kickstarter, Indiegogo) provide market validation before mass production; and the rise of “hardware-enabled SaaS” business models means hardware can generate recurring revenue rather than one-time sales.
Space tech is the most visible hardware success story. SpaceX alone is valued at roughly $200+ billion (secondary market estimates, as the company remains private) — making it the most valuable VC-backed company in history, software or otherwise. The company’s Starlink satellite internet business generates an estimated $5-6 billion in annual revenue with high margins. Relativity Space (3D-printed rockets), Astranis (small GEO satellites), and Planet Labs (Earth imaging — went public via SPAC at $2.8B) show that space tech produces venture-scale returns when the technology works.
Robotics companies are finally achieving commercial viability. Boston Dynamics (now owned by Hyundai) has commercialized Spot (the robot dog) and Stretch (warehouse robot). Figure AI raised $675 million at a $2.6 billion valuation (with investments from OpenAI, Microsoft, NVIDIA, Jeff Bezos) for humanoid robots. The thesis: labor shortages in manufacturing, logistics, and elder care create genuine demand, and advances in AI (computer vision, reinforcement learning, language models) make robots more capable than ever.
Biotech as Venture Asset Class
Biotech investing is fundamentally different from software: timelines are longer (10-15 years from discovery to approval), failure rates are higher (90% of drugs that enter clinical trials fail), and the regulatory pathway is binary (FDA approval = victory; rejection = near-zero value). But the returns for successful drugs are enormous — $1+ billion annually for blockbuster drugs — and the public markets provide a clear exit path (IPO, often at Phase II data).
Notable biotech investments in 2024-2025: Altos Labs (anti-aging, $3 billion raised), Generate Biomedicines ($273 million), and a wave of AI-enabled drug discovery companies (discussed in detail in the AI drug discovery article). The fusion of AI and biotech is the most exciting frontier: companies that can use machine learning to identify drug targets and design molecules dramatically faster than traditional methods could fundamentally change the economics of drug development — making it more like software (faster iteration, lower failure rates) and more attractive to generalist tech investors.
The Challenges That Haven’t Changed
Deep tech investing still carries structural challenges that software doesn’t. Capital intensity: fusion energy companies need hundreds of millions before generating a dollar of revenue (ITER, the international fusion project, has cost $20+ billion over decades). Timeline mismatch: the 10-year life of a VC fund is too short for nuclear reactor development (20 years from concept to operation is common). Technical risk: the physics might simply not work — there’s no “iterate toward product-market fit” in fusion or quantum computing.
The specialized expertise requirement means deep tech funds employ PhDs — materials scientists, nuclear engineers, computational biologists — in addition to the standard MBA-consultant VC profile. This changes the economics of the fund (higher salaries, lower assets per investment professional) and the culture. The deep tech VC world looks different from the Sand Hill Road stereotype, and not just because of the lab coats.
The Thesis
The bull case for deep tech goes something like this: the easy software problems are mostly solved. We have search engines, social networks, e-commerce platforms, and SaaS tools for every business function. The hard problems — climate change, disease, energy abundance, space colonization — require breakthroughs in physics, biology, and engineering that software alone cannot provide. The companies that solve these problems will create trillions in value, and they’ll be built on the foundations being laid right now.
Whether the returns materialize at venture scale is the question. The track record is thin. But the direction of travel — more capital flowing to harder problems, on longer timelines, with more government support — is unmistakable. Deep tech is no longer a niche within venture capital. It’s a major category that’s reshaping what venture capital means.