E25: Seth Winterroth, Partner at Eclipse: Investing $10B in the Physical Economy

Episode Summary

Executive Summary: Eclipse partner Seth argues that frontier tech is moving from software into the physical world, where massive value lies in industrial, defense, logistics, robotics, and food systems. He says the firm’s edge is thesis-driven investing, deep operator experience, and focusing on engineering execution rather than science risk. He is bullish on large-model applications in physical industries, but skeptical of hype cycles like general-purpose robotics and urges disciplined customer, team, and market selection.

Main Topics: Why Eclipse exists and what frontier tech means (Priority: 5/5): Seth explains Eclipse’s origin as a response to a venture gap: Silicon Valley long ignored full-stack, physical-world companies requiring multidisciplinary engineering, operations, and manufacturing expertise. Talent migration from elite hardware/software companies (Priority: 5/5): He argues the pool of founders is expanding because experienced operators from SpaceX, Tesla, Rivian, Palantir, Waymo, and similar companies are becoming founders, while software talent seeks bigger problems than incremental SaaS. Engineering execution vs. science risk (Priority: 5/5): A major investing principle is to avoid underwriting net-new invention timelines, and instead back teams that can execute complex engineering to productize known technical capabilities. Thesis-driven company building and incubations (Priority: 4/5): Eclipse runs internal thesis meetings, pressure-tests ideas, and sometimes incubates companies directly. Seth says the best opportunities come when the firm has done the work before meeting a founder. Go-to-market realities in defense and industrial markets (Priority: 4/5): He details how difficult it is to sell into defense and industrial customers, emphasizing champions, procurement, transparency, and the importance of founders being hands-on early on. AI, robotics, and physical-world infrastructure (Priority: 5/5): Seth is excited about foundation models entering the physical world, but skeptical of claims that robotics is about to get a single ChatGPT-like breakthrough; he sees tooling and infrastructure as major opportunities. Re-industrialization, supply chains, and food systems (Priority: 3/5): He connects national security, supply-chain resilience, and food quality to the broader reindustrialization movement, arguing that market demand and government support could reshape these sectors.

Key Arguments: The physical world represents a huge, under-digitized opportunity set because around 80% of global GDP still exists there. Eclipse was founded because early-stage venture largely refused to lead complex, hardware-heavy, full-stack companies; that gap is now becoming a category. Founders from SpaceX, Tesla, Rivian, Palantir, Waymo, and similar firms are stronger because they are often first-time founders but not first-time operators. Great software talent is moving toward physical-world startups because the long tail of SaaS feels incremental and less meaningful. The correct underwriting lens is engineering execution, not scientific invention timelines, because science risk is too uncertain to predict. In industrial and defense markets, customers primarily want reliability, safety, and throughput; the product must work, not just demo well. Successful pilots depend on honest communication, root-cause analysis, and selecting design partners with real budget and influence. Defense GTM is uniquely difficult and often requires founder-led selling, upstream procurement shaping, and multiple internal champions. Robotics is not about to experience a simple one-model breakthrough; it is still an iterative, systems-level grind. Developer tools and infrastructure for applied robotics will become a major category as physical AI matures. Food production and supply chains are likely to be rethought due to soil degradation, nutrient concerns, processing dependence, and security risks. Reindustrialization will likely be driven by both market demand and selective government support rather than blanket intervention.

Data Points: Years Eclipse has been operating: 10+ years - Seth says Eclipse started just over a decade ago to back physical-world frontier tech. Share of global GDP in the physical world: 80% - He uses this to frame the size of the opportunity in industrial and physical-world sectors. Internal thesis meetings per week: 2 - Eclipse runs two weekly meetings dedicated to thesis development and discussion. Net-new theses generated annually: 60+ - He says the firm generates at least one, sometimes two, new theses per week. Thesis development stages: 1.0 / 2.0 / 3.0 - Eclipse categorizes the progression of ideas internally by maturity. Time from Eclipse seed to Oxide’s first true production customers: ~5 years - Eclipse led Oxide’s first two rounds in 2019; production customers arrived last year. Capital invested into Eclipse-built seed-stage companies: 20%–25% - He says this share of recent fund capital has gone into companies built out of Eclipse. Mitra formation timeline: Summer 2022 - Eclipse built Mitra out of the firm and led the seed round that summer. Typical pilot learning cycle: 9–16 months - He says early pilots can take roughly a year or more to learn and iterate. Foundation-model scale referenced for robotics: Many billions to tens of billions of parameters - He notes some physical-world tasks may require very large foundation models, though not all robotics problems do. Defense-tech company raise range discussed: $5M–$10M - He says at this stage, founder-led selling is usually necessary because a company cannot hire its way into procurement influence.

Pivotal Quotes: "Hard is our moat." — Seth: He cites this Bezos idea to explain why Eclipse favors difficult engineering and operational complexity. "What I want to foster inside of our organization is a culture where everything's on the table. Everything can be discussed. And then the work begins." — Seth: Describing Eclipse’s thesis culture and openness to contrarian ideas. "I think a lot of great founders are looking around and say there's big problems. If we solve those problems, there's tremendous value to capture." — Seth: On why founders are moving toward physical-world industries and hard-tech opportunities.

Implications: Physical-world innovation is becoming investable at scale, but success will depend on disciplined execution, patient capital, and strong operator-founder networks. The next big winners may come from robotics tooling, defense, logistics, energy, and food-system reengineering rather than pure model hype.