Scale AI CEO on Meta’s $14B deal, scaling Uber Eats to $80B, & what frontier labs are building next
Scale AI CEO Jason Droege explains Meta’s $14B non-voting stake, why expert PhDs now do the labeling, and why survival beats aggression in a hype cycle.
- Meta invested $14B for 49% non-voting stock in Scale AI; no board seat, no preferential data access, only ~15 people transferred to Meta.
- Scale AI now has two business units each generating hundreds of millions in revenue, and has grown every month since the Meta deal.
- 80% of Scale’s expert labeler network holds a bachelor’s degree or higher; 15% hold PhDs earning significant income per task.
- Labeling tasks have shifted from simple preference ranking 18 months ago to multi-hour expert work like building full websites or explaining nuanced cancer topics to models.
- Most enterprise data is useless for AI training — the real bottleneck is digitizing human judgment specific to each organization’s culture and incentives.
- Scale signed two separate $100M government contracts in a single month; federal and international government business is growing rapidly.
- Droege’s core hiring filter: curious problem-solver, humble cross-functional collaborator, good leader — same three regardless of domain or AI disruption.
- His “not losing is a precursor to winning” philosophy argues survival through hype cycles beats high-risk bets, since customers will still need solutions in 5 years.
2025-10-09 · Watch on YouTube
