Alexandr Wang: Building Scale AI, Transforming Work With Agents & Competing With China

Alexandr Wang explains how Scale AI grew from a data-labeling API to a $29B defense and enterprise AI infrastructure company, and why China may have a data advantage in the AI race.

• Meta agreed to invest $14B in Scale AI at a $29B valuation; Wang will lead Meta’s new AI superintelligence lab.
• Scale started DoD defense AI work in 2020, years before the recent drone-fueled AI craze in the Pentagon.
• Scale’s Thunder Forge program with Indo-Pacific Command compresses military planning decision cycles from 72 hours to 10 minutes using AI agents.
• Wang believes China has a structural data advantage: government-run labeling centers, subsidy vouchers, college pipelines, and freedom to ignore copyright and privacy law.
• Wang attributes China’s rapid model progress partly to espionage leaking hyperparameter and training ‘secrets’ from US frontier labs.
• Humanity’s Last Exam launched with top models scoring 7-8%; best models now exceed 20% — benchmark saturation expected.
• Wang’s thesis on future work: the terminal state of the economy is large-scale humans managing agents, not humans being replaced entirely.
• Scale’s agentic applications business is now growing faster than its data-labeling core and is, by Wang’s account, one of the largest AI application businesses in the industry.

2025-06-18 · Watch on YouTube