How to Think About AI Lab Economics
Key Info
AI lab economics can be reduced to whether GPUs can be bought, software run, and the results sold at a profit — plus how scale changes costs and pricing power, and how much margin you can give up while still growing rapidly.
Highlights
- Start with unit economics: acquire GPUs, run software on them, and sell the output above cost.
- Scale changes both costs and pricing power, so a reported 80% gross margin should be understood in context, not in isolation.
- Faster growth may justify eating some margin — for example, paying 20% to channel partners like AWS and Google in exchange for additional capacity and demand.