Executive Summary
Ben Horowitz identifies a structural shift in AI that creates unprecedented winner-take-all dynamics through researcher scarcity. Only 40 people globally can build large-scale AI models, having worked at Google, Facebook, OpenAI, or Anthropic. This scarcity explains billion-dollar researcher valuations that appear 'absolutely bananas' from the outside. Unlike previous technology cycles requiring infrastructure buildout, AI deployment leverages existing internet infrastructure, accelerating adoption timelines to 12-24 months. Horowitz argues this creates a new physics of company building where capital can solve problems previously constrained by engineering time - a fundamental departure from software's historical 'you cannot throw money at the problem' rule. The implications extend beyond venture capital into national competitiveness, where America's entrepreneurship culture and regulatory environment position it to maintain technological leadership. However, policy risks remain the primary threat, with Horowitz citing the Biden administration's GPU approval requirements as an example of how quickly regulatory overreach could eliminate America's chip advantage. The convergence of AI researcher scarcity, capital abundance, and infrastructure readiness creates an environment where market sizes could reach $5 trillion rather than traditional $50 billion assumptions, fundamentally altering valuation frameworks.
Key Insights
what Ben Horowitz said“If you haven't been at Google or Facebook or OpenAI or Anthropic and somebody gave you hundreds of millions of dollars to try and build a giant model and you weren't one of the main people, then you probably don't know how to do it... what if there are only 40 of them in the world? Then it kind of changes the math on it a little bit.”
This is a preview. Log in to see the full analysis including investment opportunities, risks, catalysts, and detailed insights.