Ben Horowitz & Marc Andreessen: Why Silicon Valley Turned Against Defense (And How We’re Fixing It)

Marc and Ben recount how Silicon Valley was deeply integrated with defense from the 1950s through the 1990s—Netscape sold to the Pentagon in 1994, both founders had top-secret clearances. Post-Vietnam and post-Iraq sentiment, combined with heated 2010s politics, created escalating hostility culminating in Google's Maven revolt where employees forced the company to abandon a Pentagon AI contract.

• •Netscape's first Pentagon sales call was 1994; defense work was considered natural and patriotic through the 1990s
• •Google Maven project (late 2010s) marked peak hostility: employee revolt forced Google to cancel Pentagon AI contract
• •Post-Vietnam and post-Iraq politics created moral framing where working with military was seen as unethical
• •Universities like Stanford divested military R&D into separate institutes; many banned ROTC and intelligence recruiting
• •Hostility was reciprocated: Washington developed its own contempt for Silicon Valley during this period

Katherine Boyle contrasts the Bob Noyce generation (hands-on engineers from rural America building physical things) with the Harvard dorm room culture popularized by The Social Network. Geographic and cultural distance matters: DC sees uniformed service members daily on the Metro; Silicon Valley can go decades without seeing anyone in uniform. The 2020 'It's Time to Build' essay marked the turning point.

• •Bob Noyce came from rural Iowa; Marc from rural Wisconsin—both knew servicemembers personally, unlike coastal tech workers
• •The Social Network (2010) defined a generation: Harvard dorm room coding vs. working with your hands in factories
• •Washington DC's daily reality includes Pentagon Metro stops and defense contractors; Silicon Valley has zero military presence
• •Marc's 'It's Time to Build' essay (2020) during COVID exposed America's inability to manufacture PPE, sparking rethink
• •Palmer Luckey's arc—fired from Meta for political donations, now defense hero—symbolizes the cultural reversal

David Ulevitch shares the origin story of American dynamism: meeting Palmer Luckey in 2019, competing with Katherine Boyle (then at General Catalyst) for Anduril's Series B, having to split the deal, and awkwardly asking for her diligence notes. Garrett Langley of Flock Safety later played matchmaker, suggesting they join forces at a16z, which led to the 2021 launch of the American dynamism practice.

• •David led Anduril Series B in 2019 after flying to Costa Mesa and refusing to leave until Palmer signed
• •Katherine had done all the diligence work; David asked for her notes after forcing the split—'one of the more embarrassing' moments
• •Flock Safety's Garrett Langley called Katherine at 11:30pm saying David 'wouldn't let me leave the restaurant'
• •Garrett suggested Katherine join a16z so they could work together; David called two minutes later to propose dinner
• •American dynamism fund officially launched in 2021 with full firm backing and dedicated capital

David explains why hardware is no longer the capital-intensive nightmare investors feared: early American dynamism companies used commodity hardware (Flock's cameras were phone cameras in boxes) paired with advanced software for computer vision and autonomy. Government sales cycles are predictable, not Christmas-dependent. Equipment financing is available. The moment arrived when software could transform physical products.

• •Flock Safety started with commodity cell phone cameras in weatherproof boxes with LTE—not custom hardware
• •Anduril's early Sentry Tower used off-the-shelf optics paired with advanced computer vision software
• •Government/enterprise sales are predictable and manageable, unlike consumer electronics' Christmas-or-bust cycles
• •Equipment financing available for large-scale hardware projects reduces venture capital burden
• •Key insight: advanced software (AI, computer vision, autonomy) now makes commodity hardware incredibly powerful

David outlines the energy investment thesis: society has insatiable demand for power (AI compute, EVs, aging grid). Historical data shows pouring energy into economies creates growth. The firm needs modular, mobile, transmittable power generation—hence investments in Exawatt (solar) and Radiant Nuclear. Three companies are now on China's 'unreliable entities list' and banned from buying Chinese batteries.

• •AI compute loads, electric vehicles, and grid electrification create unprecedented power demand
• •Historical pattern: more energy input = economic growth across all modern economies
• •Investment focus: generation (nuclear, solar), transmission infrastructure, and storage (batteries)
• •Exawatt and Radiant Nuclear represent modular, deployable power generation systems
• •Three portfolio companies banned from Chinese batteries—critical need for domestic battery manufacturing

Katherine describes how Space 1.0 required vertical integration (build everything yourself, take 10 years to orbit). Space 2.0 companies like Apex (satellite buses) and Northwood (ground stations) are deconstructing the stack, reaching orbit in 13 months by focusing on specific components. SpaceX's manufacturing lessons are spreading across tens of thousands of engineers who've left to build new companies.

• •Apex Space reached orbit in 13 months from clean sheet design—fastest in history—by only building satellite buses
• •Space 2.0 companies work with legacy primes and new players, not competing but enabling them
• •Northwood Space tackles ground stations—critical bottleneck as LEO saturates with 85% of mass to orbit from SpaceX
• •SpaceX's real legacy: tens of thousands of engineers learned manufacturing processes now spreading across industries
• •Unbundling vertical integration allows faster iteration and specialization vs. Space 1.0's build-everything approach

Katherine explains the golden triangle: desperate customers (Pentagon + Congress), abundant downstream capital, and founders graduating from SpaceX/Anduril with manufacturing expertise. SpaceX and Palantir 'walked so others could run'—both had to sue the government in the 2000s. Now legacy primes actively partner with startups, and procurement reform is finally happening. Defense 2.0 companies grow faster by working with incumbents, not just competing.

• •SpaceX and Palantir (founded 2002-2003) had to sue the government to break in; they paved the way for everyone else
• •2025 NDAA includes procurement reforms that will benefit entire defense ecosystem, not just startups
• •Downstream capital now abundant—the 'messy middle' funding gap that killed companies no longer exists
• •Founders leaving Anduril/SpaceX with blessing to build adjacent companies (e.g., Base Power's founder led Anduril manufacturing)
• •Defense 2.0 companies partner with legacy primes rather than only competing—faster growth, richer customer set

Katherine and David discuss how Ukraine proved $1,000 drones can destroy $100M tanks, changing war economics forever. The concept of 'attritable systems'—cheap, expendable hardware you can lose—is now central to Pentagon thinking. China supplied both sides of Ukraine, building up their industrial base while America's atrophied. Battlefield iteration cycles are now days, not decades. Wars of mass production favor whoever can manufacture fastest.

• •Ukraine demonstrated $1,000 drones destroying $100M exquisite systems—fundamentally changed war economics
• •Pentagon now focused on 'attritable systems': cheap, expendable hardware you can afford to lose at scale
• •China supplied both sides in Ukraine, building manufacturing capacity while US industrial base remained stagnant
• •Battlefield iteration cycles in Ukraine are days/weeks, not the five-year Pentagon planning cycles
• •America exported manufacturing for 40 years; now can't produce attritable systems at scale like China can

Marc argues America won't get 1980s factory jobs back—and shouldn't want them. The future is e-bikes (not bicycles), Tesla-style automated plants, and 'blue collar plus' jobs maintaining sophisticated robotics. Products are more complex (batteries, chips, computers), manufacturing is more automated, and jobs are higher-skilled and better-paid. China's set up to build tomorrow's robots by default; America needs to re-industrialize for future products, not past ones.

• •Heraclitus: 'Man cannot step in same river twice'—even with perfect policy, 1980s factory jobs aren't coming back
• •Future manufacturing: e-bikes with batteries/computers, not simple bicycles; Tesla factories, not assembly lines
• •Jobs will be 'blue collar plus': higher-skilled, higher-paid, maintaining sophisticated automation and robotics
• •China currently positioned to dominate robotics manufacturing—the biggest industry ever built—by default
• •Opportunity: re-industrialize America for future products (AI robots, advanced hardware) not past products (simple goods)

Marc recounts how 1980s economics textbooks taught communism was superior to capitalism due to centralized efficiency. The five-year plan—invented by Stalin—still infects corporate America and the Pentagon. China's centralized system has advantages for pumping out billions of phones, but America won Cold War I through dynamism, creativity, and entrepreneurship. The answer isn't more state control; it's unleashing even more American innovation and flexibility.

• •1980s undergraduate textbooks stated as fact that communism would outperform capitalism economically
• •Five-year plans invented by Stalin in 1920s; still used by corporations and Pentagon despite routinely failing
• •China's centralized approach has advantages for mass production, but lacks innovation and adaptation speed
• •America won Cold War I by being more American—dynamic, entrepreneurial, creative—not by copying Soviets
• •NYC mayor proposing government-run grocery stores shows the centralization debate is back and must be re-fought

Katherine explains how the Department of Defense still operates on Soviet-style five-year planning cycles from the 1960s-70s, while Ukraine iterates in days. Current RFPs require 'past performance'—favoring incumbents regardless of budget overruns. Reforms in the 2025 NDAA will shift toward 'what can you deliver today?' enabling startups. Operation Warp Speed showed capitalism at its best: government set demand signal, private companies competed, three vaccines in record time.

• •Pentagon procurement still uses five-year cycles and ten-year production timelines from 1960s-70s regulations
• •Current RFPs require past performance regardless of quality—locks out startups, favors incumbents with overruns
• •2025 NDAA procurement reforms shift to 'what can you deliver today' vs. 'what did you do 20 years ago'
• •Operation Warp Speed model: government distributes, private companies compete on innovation—three vaccines in record time
• •Ukraine proves modern battlefield requires days/weeks iteration, not decades—Pentagon must match that speed

David and Katherine outline expansion into 'shifting left'—the supply chain components needed before you build anything. Mining, critical minerals, magnets, motors, copper, steel represent trillions in economic spend where software hasn't penetrated. Advanced algorithms can optimize mining (pressure, temperature, site selection). Manufacturing and supply chain resilience are critical for conflict scenarios. These legacy categories are ripe for technology transformation.

• •Shifting left to supply chain: mining, critical minerals, magnets, motors, copper, steel—trillions in economic value
• •Software has barely touched mining despite complex optimization opportunities (pressure, temperature, site selection)
• •Manufacturing and supply chain resilience critical for conflict scenarios—can't rely on adversary supply chains
• •Talent now shifting attention to these 'legacy' categories as technology makes them venture-scale opportunities
• •Hadrian (factory automation) early example; many more opportunities to automate production of 'dumb parts' at scale

Katherine describes Odyssey's thesis around Education Savings Accounts (ESAs)—state programs letting parents opt out of public schools and receive cash for alternative education. Texas passed universal ESA; other states following. COVID accelerated homeschooling and hybrid models. Odyssey builds financial rails to track and regulate ESA spending. AI-powered one-on-one tutoring and alternative learning methods are transforming K-12 education.

• •Education Savings Accounts let parents opt out of public schools, receive state cash for alternative education
• •Texas passed universal ESA; trend spreading as parents want more control over children's learning
• •COVID normalized homeschooling and hybrid learning models, creating demand for supporting technology
• •Odyssey provides financial rails to track ESA spending and ensure compliance with state regulations
• •AI-powered one-on-one tutoring can support special needs and different learning capacities at scale

Katherine describes what makes American dynamism founders different: they're built across America (Austin, Atlanta), not just SF. Many served in military (Saronic's Dino: 11 years Navy SEALs before business school). They have clearances before starting companies, deeply understand customer needs, and speak the government's language. This isn't Harvard dorm room coding—it's founders who've lived the problems they're solving.

• •American dynamism companies built across America: Saronic (Austin), Flock (Atlanta)—not concentrated in SF/SV
• •Founders often have military service background (Saronic's Dino: 11 years Navy SEALs enlisted before founding)
• •Many have security clearances before starting companies—understand government customer from inside
• •Key screening criteria: how deeply does founder understand the customer they're selling to?
• •Not 'Harvard dorm room' archetype—these are operators who've lived in the problem space they're solving