Key Takeaways
- • Venture capital now funds defense hardware before any government contract exists, with Erin Price-Wright and Connor Love of Andreessen Horowitz (a16z) arguing that risk dollars entering the defense ecosystem in the first half of 2026 exceeded all of the prior year combined.
- • Castelion spent roughly two and a half to three years developing its first hypersonic missile on private venture capital before government contracts followed, illustrating a reversal of the traditional requirement-then-prime procurement model.
- • A portfolio company's printed circuit board pass rate rose from roughly 50% to 70-90% after switching to new U.S. PCB suppliers and co-developing designs with its end customer.
- • Price-Wright argues the binding constraint on the American industrial buildout is political will and community acceptance rather than technology, and that this question remains unresolved.
- • a16z's 2026 American Dynamism 50 list maps the defense industrial stack across inputs (e.g., Mariana Minerals), factories (e.g., Hadrian), and end systems (e.g., Anduril, Castelion).
Summary
According to Erin Price-Wright and Connor Love, who lead the American Dynamism practice at Andreessen Horowitz, the traditional sequence in U.S. weapons development has inverted. Historically, the government wrote a requirement and prime contractors waited for funding. Today, venture capital funds hypersonic missiles, mines, and factories years before a government contract exists. Price-Wright says more risk dollars entered the defense ecosystem in the first half of 2026 than in all of the prior year, a year that was itself many times the one before. Love points to JPMorgan and large credit firms financing physical-world projects, suggesting the capital base is widening beyond venture funds.
Castelion is the central example. According to the discussion, the company spent roughly two and a half to three years developing its first hypersonic missile on private capital before government contracts arrived. Love credits the company's refusal to pursue multiple products, including small drones, with building process knowledge, supplier relationships, and design-for-manufacturing capability. He described the missile as "not a super exquisite thing, but it's really hard to build a missile that goes that fast and is reliable."
Price-Wright and Love also describe an emerging "manufacturing middle" of suppliers that co-engineer parts rather than passively building to decades-old specifications. Love illustrates the issue with Boeing's 737 and Honeywell's landing gear, noting that sub-tier suppliers have been in "receive mode" for roughly 40 years. One portfolio company reportedly ordered twice the printed circuit boards it needed three years ago, expecting half to fail quality testing, and half did. After switching to new U.S. PCB suppliers, 70-90% now pass testing, which Price-Wright attributes to close co-location and co-development with the customer.
The a16z framework groups the industrial stack into inputs (Mariana Minerals, which bought a copper mine to control its value chain), factories (Hadrian, which Love says requires addressing "15 or 20 different things"), and end systems (Anduril, Castelion, Seronic, Joby, and Archer). Love argues the same actuators, motors, and valves used in military hardware are central to future robotics, and cites DJI's trajectory from consumer drones to robot production as an example.
Price-Wright identifies political will rather than engineering as the main constraint on U.S. power and industrial expansion. She said, "I don't think that is a technology problem. I think that's a we need to do some soul-searching about our will as an American people and do we want these projects to be built." Love, by contrast, argues that once the demand cycle begins, the ratchet effect makes reversal financially unlikely, even if deliveries fall short of targets.
Castelion is the central example. According to the discussion, the company spent roughly two and a half to three years developing its first hypersonic missile on private capital before government contracts arrived. Love credits the company's refusal to pursue multiple products, including small drones, with building process knowledge, supplier relationships, and design-for-manufacturing capability. He described the missile as "not a super exquisite thing, but it's really hard to build a missile that goes that fast and is reliable."
Price-Wright and Love also describe an emerging "manufacturing middle" of suppliers that co-engineer parts rather than passively building to decades-old specifications. Love illustrates the issue with Boeing's 737 and Honeywell's landing gear, noting that sub-tier suppliers have been in "receive mode" for roughly 40 years. One portfolio company reportedly ordered twice the printed circuit boards it needed three years ago, expecting half to fail quality testing, and half did. After switching to new U.S. PCB suppliers, 70-90% now pass testing, which Price-Wright attributes to close co-location and co-development with the customer.
The a16z framework groups the industrial stack into inputs (Mariana Minerals, which bought a copper mine to control its value chain), factories (Hadrian, which Love says requires addressing "15 or 20 different things"), and end systems (Anduril, Castelion, Seronic, Joby, and Archer). Love argues the same actuators, motors, and valves used in military hardware are central to future robotics, and cites DJI's trajectory from consumer drones to robot production as an example.
Price-Wright identifies political will rather than engineering as the main constraint on U.S. power and industrial expansion. She said, "I don't think that is a technology problem. I think that's a we need to do some soul-searching about our will as an American people and do we want these projects to be built." Love, by contrast, argues that once the demand cycle begins, the ratchet effect makes reversal financially unlikely, even if deliveries fall short of targets.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.