We investigate four constraints to scaling AI training: power, chip manufacturing, data, and latency. We predict 2e29 FLOP runs will be feasible by 2030.| Epoch AI
Scaling will run out. The question is when.| www.aisnakeoil.com
How big is YouTube? It's not easy to find out, but we did two years of research and figured out how...| Ethan Zuckerman