‘Nothing Ever Happens’ Is Over

Naval • • 19min → 3 min • #23
‘Nothing Ever Happens’ Is Over
Watch on YouTube

Summary

  • Naval Ravikant and Nivi discuss how AI is reshaping startups, warfare, biology, and hardware, arguing we’ve entered a period of rapid, irreversible change where sci-fi technologies are becoming real and the old “nothing ever happens” era is over.
    • AI in startups: small teams, no tools, fully interconnected graphs
      • At Impossible, Naval’s current company, they avoid traditional organizational tools like Slack and project management software, relying instead on a flat, hub-and-spoke structure where everyone communicates directly through the CEO.
      • Naval prefers a “fully interconnected graph” model where every team member is highly intelligent and self-directing, navigating to whoever they need to solve a problem, rather than a hierarchical tree structure that creates politics and bottlenecks.
      • AI is used implicitly rather than as an explicit organizational tool: it summarizes complex code and papers, maps expertise across the codebase, and can generate on-demand Gantt charts, dashboards, and status reports by analyzing code, designs, vendor documents, and even company email.
      • AI also acts as a cross-functional force multiplier: hardware engineers can write enough software to bring up new devices without waiting for software teams, and AI researchers can build their own test harnesses, reducing the need for rigid APIs and making everyone more generalist.
    • The AI industry: centralization, AGI uncertainty, and open questions
      • The AI landscape is dominated by roughly four to five companies (including Nvidia for hardware), and the big open question is whether this consolidates further into a monopoly/oligopoly or fragments.
      • Key unknowns include whether distributed training is possible (current consensus says no), whether models will run out of data, and whether AGI is achievable.
      • Naval is skeptical of claims about “world models,” distinguishing between visually impressive generative environments and true world models that let agents predict consequences of actions and adjust behavior through reinforcement learning loops.
      • He notes that post-COVID, the world is changing much faster geopolitically, economically, and technologically, and VCs are now funding hardware, rockets, drones, and other “sci-fi” technologies.
    • Drones and the democratization of violence
      • Drones are still under-leveraged and will continue to transform warfare because attackers have advantages in kinetic energy and surprise, while defenders are spread thin.
      • Naval draws a historical parallel: the rifle enabled the fall of feudal states and rise of nation-states; nuclear weapons consolidated power among seven to nine sovereign nations; drones bring mutually assured destruction down to the individual level.
      • The open question is whether drone power centralizes under a few large states or becomes so democratized that any individual can be deadly.
    • Biothreats and the democratization of biological weapons
      • AI dramatically lowers the barrier to creating biological weapons, just as vibe coding democratized software development, potentially increasing the number of people who could access such weapons by hundreds of thousands of times.
      • The counterargument is that AI can also accelerate vaccine and therapy research, but Naval worries that medical regulations and bureaucracy prevent rapid response even in emergencies, as seen with COVID vaccine timelines.
      • He suggests anonymized, open health data and “right to try” frameworks could enable better defenses, but fears these will only emerge reactively during crises.
    • Hardware renaissance unlocked by AI
      • Historically, great hardware was often crippled by terrible software; Apple succeeded by excelling at both.
      • AI changes this: companies that are strong in hardware but weak in software can now rely on AI agents to interface with hardware directly, or on individuals using tools like vibe coding to build software quickly.
      • This dynamic benefits China (dominant in consumer electronics manufacturing) and Nvidia (wants maximum GPU adoption), both of whom have strong incentives to push open-source AI models that commoditize software and unlock more hardware.
    • Optimism as a creative act
      • Naval argues that doom scenarios are easier to imagine than positive futures because optimism requires creativity, while pessimism only requires extrapolating current trends.
      • Historically, predicted catastrophes (environmental collapse, world wars) have not materialized in the ways feared, and entirely new categories of jobs and technologies have emerged that no one could have predicted.
      • He advocates for deliberate, even irrational optimism, arguing that pessimism is unproductive and that optimists are the people you want in a crisis.
Back to Naval