Sean Liu of Founders Fund on evaluating technical risk in hard tech — and why most VC frameworks don't apply

Summary

Sean Liu, a partner at Founders Fund with prior operating experience as CFO/COO at Solugen, offers a framework for hard tech investing that cuts against how most venture firms approach the category.

Technical Risk Is a Spectrum, Not a Binary

The first question Liu applies to any hard tech deal is whether the company needs to invent new science. If the answer is yes, the investment profile changes dramatically. Founders Fund positions itself as a backer of revolutionary technology, but Liu is explicit that the science-to-company translation is frequently missing from founder pitches. The firm's preference is for companies making incremental improvements to scale proven science, not those requiring multiple new scientific fields to converge simultaneously.

Liu frames thesis risk using what he calls an "and statement" model. Every codependent assumption multiplies downside probability. He draws a direct analogy to semiconductor manufacturing: a 7,000-step process with 0.1% yield loss at each step produces zero output. The same compounding logic applies to hard tech investment theses. Founders Fund targets companies with one or two core "need to believes," not five or six.

How Founders Fund Differentiates Hard Tech from Capital Intensity

Liu draws a clean distinction between hard tech and capital-intensive businesses, noting they are not the same thing. The firm's investment in Crusoe last year was framed primarily as an energy infrastructure and long-term US power gap thesis, not a pure hard tech bet. By contrast, portfolio companies like Neuralink and SpaceX represent genuine technical risk where the science itself is the moat.

For growth-stage deals, the key variables shift to supply chain readiness, upstream input availability, and whether ecosystem partners can scale alongside the company. Liu notes that hard tech founders routinely oversell technology maturity while underestimating the commercial build required. The SpaceX example is instructive: it entered an existing launch market and captured share rapidly rather than waiting for a market to materialize.

Hard Tech Power Laws Are More Extreme Than Software

Liu argues hard tech is more power-law concentrated than software, not less. SpaceX effectively is the space launch market. Anduro dominates defense mining. Large pharma companies are massive outliers relative to everything else in the sector. The structural reason is capital advantage: incumbents like DuPont carry hundreds of billions in deployed capex, making marginal production costs low enough that displacement becomes nearly impossible. Fewer winners emerge, but the defensibility of those winners is higher than in software.

Clean Tech 1.0 as a Cautionary Framework

Liu characterizes the clean tech investment wave of the late 2000s and early 2010s as a "martyr" category, borrowing a framing from Garb that being one step early makes you a genius and two steps early makes you a martyr. The industrial base was not ready, polysilicon processing capabilities were insufficient, and utilities lacked the ability to manage non-firm power systems. The underlying economics required government subsidies assumed to last for decades, which did not hold.

The category definition problem compounded the timing problem. Tesla was classified as an automotive company rather than a clean tech company, meaning clean tech funds missed the category's dominant power law winner. Liu draws a direct parallel to the present, arguing that nuclear should be understood as the cleanest available technology, and that the right lens for hard tech investing is commercial pull clarity, not technology taxonomy.

Solugen Operating Experience as Due Diligence Foundation

Liu's time at Solugen during COVID shaped his view that hard tech CFO and COO work is fundamentally different from SaaS finance. He personally counted inventory after firing the head of FP&A within two weeks of joining. His core observation on chemical startups applies broadly: no new chemical company at scale has emerged in 30 to 40 years, partly because R&D is centralized among petrochemical giants with strong incentives to act as patent trolls, and partly because factory construction starts at roughly $300 million on the low end, creating a structural valley of death that standard venture frameworks do not adequately price.

On IP strategy, Liu's view is that patents are necessary but insufficient. Solugen's defensibility rested on broad umbrella patents combined with trade secrets and the specificity of its enzyme-based chemistry platform, which made replication difficult even where the formulation direction was publicly known.