Why AI Labs Are Starting to Look Like Sports Teams

• AI infra spend: $840B deployed; revenue still disproportionately small.
  • OpenAI leads at >$10B annualized revenue.
• Three PMF signals: coding ($3B run-rate), reasoning, ChatGPT as daily habit.
• Compute abundance shifts the moat from scale to talent.
  • DeepSeek proved frontier research possible without massive compute.
• Star researchers now command athlete-scale comp: hundreds of millions to billions.
• Labs = sports franchises: wealthy backers, short contracts, poaching wars.
  • Unstable at team level; stable at ecosystem level.
• Market calming: five-player oligopoly consolidating into known competitive structure.

X discourse

• @signulll: “big tech structurally can’t hire the ppl who actually get ai… those ppl are founding companies” (788 likes)
• @ajratner: “frontier data lab & partner to leading AI labs — more research vectors than we can possibly pursue” (471 likes)
• @hijunedkhatri: “hot take: ai talent will be paid more than professional athletes” (190 likes)
• @MarcoFigueroa: “AI companies have turned talent recruitment into a spectator sport unveiling new signings with same fanfare as NBA” (2 likes)
• @Arnav_ct: “every top AI lab is functioning like an F1 team with talent exodus as looming threat at all times” (0 likes)

David Cahn, Sequoia Capital · 2025-06-17 · Read on sequoiacap.com